sphingomyelin ratio and lamellar body count for fetal lung maturity: a meta-analysis

sphingomyelin ratio and lamellar body count for fetal lung maturity: a meta-analysis

European Journal of Obstetrics & Gynecology and Reproductive Biology 169 (2013) 177–183 Contents lists available at SciVerse ScienceDirect European ...

608KB Sizes 0 Downloads 77 Views

European Journal of Obstetrics & Gynecology and Reproductive Biology 169 (2013) 177–183

Contents lists available at SciVerse ScienceDirect

European Journal of Obstetrics & Gynecology and Reproductive Biology journal homepage: www.elsevier.com/locate/ejogrb

Lecithin/sphingomyelin ratio and lamellar body count for fetal lung maturity: a meta-analysis Anouk E. Besnard a, Soetinah A.M. Wirjosoekarto b,*, Kimiko A. Broeze c, Brent C. Opmeer d, Ben Willem J. Mol c a

Faculty of Medicine, University of Amsterdam, Amsterdam, The Netherlands Department of Obstetrics and Gynecology, Maastricht University Medical Centre, Maastricht, The Netherlands Centre for Reproductive Medicine, Department of Obstetrics and Gynecology, Academic Medical Centre, Amsterdam, The Netherlands d Clinical Research Unit, Academic Medical Centre, Amsterdam, The Netherlands b c

A R T I C L E I N F O

A B S T R A C T

Article history: Received 14 September 2012 Received in revised form 23 January 2013 Accepted 9 February 2013

Objective: To determine and compare the diagnostic accuracy of the lecithin/sphingomyelin (L/S) ratio and lamellar body count (LBC) in the prediction of neonatal respiratory distress syndrome (RDS). Study design: A systematic review was performed to identify studies comparing either the L/S ratio or the LBC with the occurrence of RDS published between January 1999 and February 2009. Two independent reviewers performed study selection and data extraction. For each study sensitivity and specificity were calculated. Summary receiver-operating characteristics (ROC) curves, assessing the diagnostic performance of both tests, were constructed. A subgroup analysis was performed to estimate the sensitivity and specificity of the various cut-off values. Results: 13 studies were included. The ROC curves of the collected data illustrate that the LBC and L/S ratio perform equally well in the prediction of RDS. Comparison of the two summary ROC curves of each test indicates that the diagnostic performance of LBC might even have a slight advantage over L/S ratio. Due to the wide cut-off range it was not possible to define specific cut-off values with the best accuracy. Conclusion: We recommend replacing the L/S ratio as gold standard with the lamellar body count since the LBC is easy to perform, rapid, inexpensive, and available to all hospitals 24 h per day. ß 2013 Elsevier Ireland Ltd. All rights reserved.

Keywords: Fetal lung maturity Lamellar body count Lecithin/sphingomyelin ratio Neonatal respiratory distress syndrome

1. Introduction Respiratory distress syndrome (RDS) is a major cause of neonatal morbidity and mortality, affecting approximately 1% of all live births and 10% of all preterm infants [1]. It is caused by insufficient production of surfactant by type II pneumocytes, along with structural immaturity of the lung. The risk and severity rise with increasing prematurity, and infants born before 29 weeks of gestation have a 60% chance of developing RDS [2]. RDS may be prevented with antenatal steroid therapy and prophylactic (early) administration of exogenous surfactant [3]. In management strategies to limit the risk of RDS, the assessment of fetal lung maturity (FLM) in amniotic fluid can assist in determining the timing of delivery, particularly in pregnancies with maternal and/or fetal complications for which a temporizing

* Corresponding author at: Department of Obstetrics and Gynecology, Maastricht University Medical Centre, P.O. Box 5800, 6202 AZ Maastricht, The Netherlands. E-mail address: [email protected] (Soetinah A.M. Wirjosoekarto). 0301-2115/$ – see front matter ß 2013 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ejogrb.2013.02.013

strategy in favor of fetal lung maturation is still considered safe, but which may eventually require preterm delivery [4]. Measurement of the lecithin/sphingomyelin (L/S) ratio in amniotic fluid by thin-layer chromatography for the prenatal prediction of FLM was first introduced in 1971 by Gluck et al. From 30 weeks of gestation onwards, the concentration of lecithin begins to increase significantly, while the sphingomyelin concentration remains approximately the same. The L/S ratio has remained the gold standard of FLM testing in the neonate, with 2.0 as a commonly accepted cut-off value, above which the risk for RDS is low, and which will normally be reached at a gestational age of 35 weeks [5]. The L/S ratio is a technically difficult test that requires trained personal to interpret. It is time-consuming, costly, prone to subjective interpretation, not universally available nor available around the clock, and it cannot be determined in fluids contaminated by blood or meconium. The lamellar body count (LBC) has been proposed as a potential replacement of the L/S ratio. Lamellar bodies represent a storage form of pulmonary surfactant within type II pneumocytes,

178

A.E. Besnard et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 169 (2013) 177–183

secretion of which increases with advancing gestational age, thus enabling prediction of the degree of FLM. The size of lamellar bodies is similar to platelets, which permits the use of widely available cell counters to quantify the lamellar bodies in amniotic fluid, as described by Dubin [6]. This makes the LBC an easy-toperform, rapid, inexpensive FLM test which is available to all hospitals 24 h per day. Previous studies have suggested using the LBC as an initial assessment prior to the use of L/S ratio [7–13]. The hypothesis that LBC is an equal or possibly better predictor for the occurrence of RDS than the gold standard, namely the L/S ratio, was tested in the meta-analysis of Wijnberger et al. [14], but data from the most recent studies were not included, and management of antenatal and neonatal care keeps developing. In the last decade there have been changes in clinical management to optimize care, such as neonatal treatment of infants at low gestational age, treatment with antenatal corticosteroids, e.g. single dose versus multiple courses, and changes in the diagnostic criteria for RDS. Since these could possibly influence the outcome of the performance of the LBC and L/S ratio, we performed an updated meta-analysis, comparing the accuracy of the LBC and the L/S ratio in the prediction of RDS.

2. Methods 2.1. Literature search and study selection A systematic literature search was performed in Medline and Embase to identify articles published between January 1999 and February 2009. Keywords used were lecithin/sphingomyelin ratio or L/S ratio or lamellar body count or LBC and respiratory distress syndrome or RDS or hyaline membrane disease or HMD. Duplicate citations were detected and removed. Cross-references were checked for additional eligible articles. Two reviewers (KAB and AEB) independently screened all identified studies by reading the title and abstract. When in doubt, the whole article was read. The final selection was made by using pre-defined inclusion and exclusion criteria. Studies were included if they reported on pregnant women at risk for preterm delivery in whom the FLM was tested by either the LBC or the L/S ratio. The outcome had to be RDS in the neonate. If the reported data were sufficient to construct a two-by-two table of the test result (LBC and L/S ratio) the study was included. Articles published in a language other than English were excluded. 2.2. Data extraction We extracted the following data: year of publication, first author, country of investigation, language of publication, total number of included patients and patients with analyzable data. Subsequently, each of the included studies was scored on the following design characteristics concurrent to the previous metaanalysis [14]: (1) sampling, (2) data collection, (3) study design, (4) blinding for the test results when RDS was diagnosed and (5) verification bias [15]. In addition, information was gathered on the following patient characteristics; minimal and maximal gestational age, inclusion of multiple pregnancies, inclusion of diabetic pregnancies, inclusion of women with ruptured membranes and use of corticosteroids. Moreover, the way amniotic fluid was collected (abdominal, vaginal or both), the time interval between amniocentesis and delivery, whether samples contaminated with blood or meconium were excluded, and how RDS was defined (clinical criteria, radiological criteria and/or criteria for oxygen therapy) were scored. Finally, the laboratory methods used to determine the L/S ratio and LBC as well as the used cut-off values were reported.

2.3. Study quality We assessed the methodological quality of the included studies using the QUADAS checklist, a tool for quality assessment of diagnostic accuracy studies [16]. Included studies were assessed on 15 items on selection, verification, description of tests and study population. 2.4. Statistical analysis For each individual study, the prevalence of RDS was calculated, as well as the sensitivity and specificity of both the L/S ratio and LBC in the prediction of RDS. Sensitivity was defined as the proportion newborns with RDS in which the test predicted immaturity of the fetal lungs, whereas specificity was defined as the proportion newborns without RDS in which the test predicted mature fetal lungs. Secondly, heterogeneity in sensitivity and specificity between studies was explored using scatter plots. Variation or heterogeneity of the results of the studies included in the meta-analysis can be the result of differences in cut-off values, bias due to flawed design, different clinical subgroups, or chance. The random-effects approach estimates and incorporates the amount of between-study variability in both sensitivity and specificity. All accuracy estimates from different studies in terms of sensitivity and specificity were plotted in receiver-operating characteristics (ROC) space. A pooled estimate for sensitivity and specificity was estimated with bivariate regression analysis, and the corresponding summary ROC (sROC) curve was constructed [17]. The bivariate regression model simultaneously estimates sensitivity and specificity within a single model, which also accommodates the inverse association between sensitivity and specificity due to threshold effects. At present, the statistical procedure to estimate the bivariate model cannot accommodate multiple data points from the same study, e.g. when a study reports sensitivity and specificity for different cut-off values. In order to evaluate accuracy measures over the whole range of reported cut-off values including all studies, we did not limit our analysis to a single cut-off value. We estimated accuracy measures for all reported cut-off values by assuming that the shift in accuracy (higher sensitivity and lower specificity) due to different cut-off values is accounted for by the correlation term, as specified in the bivariate model. Consequently, the sROC point reflects the average operating point on the curve, but as it does not reflect the accuracy for a particular cut-off, this point itself is clinically not very informative. The sROC curve corresponding with the estimated model, however, reflects the change in accuracy (sensitivity and specificity) associated with a shift in positivity threshold. In order to avoid the results being biased toward studies reporting data for multiple cut-offs, results are based on averaged model estimates from stratified bootstrap samples. 3. Results 3.1. Literature search and study selection The literature search in Medline and Embase yielded 144 articles, of which 46 were read in full text. Thirteen articles were eligible. Of these articles, one reported solely on the predictive capacity of the L/S ratio [18], nine reported solely on the predictive capacity of the LBC [19–27], and three articles reported on the predictive capacity of both tests [28–30]. 3.2. Data extraction The characteristics of the included studies are listed in Table 1. All studies were designed as cohort studies, except one which was

<72 h Yes Unknown Unknown

3.3. Study quality Table 2 shows the quality assessment with the adjusted QUADAS tool. LBC and L/S ratio were both considered as index tests, and the clinical diagnosis of RDS as the reference test. None of the studies met all criteria. 3.4. Statistical analysis

PPROM = preterm premature rupture of membranes.

Yes Winn MacMillan (2005)

No

Yes

No

Yes

Unknown

Abdominal/vaginal/cesarean

Unknown Yes Unknown Unknown Yes Roiz Hernandez (2002)

No

Yes

No

No

Unknown

Abdominal/vaginal/cesarean

<72 h <48 h Yes Yes Included Included Excluded Included Yes Yes Piazze 1 (1999) Piazze 2 (2005)

No No

Yes Yes

No No

No No

Included Included

Abdominal Adominal

<72 h No Included Yes Neerhof (2001)

No

Yes

No

No

Included

Unknown

Abdominal/vaginal

<24 h No Unknown Unknown Yes Khazardoost (2005)

No

Yes

No

No

Unknown

Unknown

<7 days <72 h <48 h Yes Yes No Included Unknown Included Included Excluded Included Yes Yes Yes Ghidini (2005) Haymond (2006) Karcher (2005)

No No No

Yes Yes Yes

No No No

No No No

Excluded Unknown Included

Abdominal/vaginal Abdominal/vaginal Abdominal

<72 h Yes Included Included Yes Chapman (2004)

No

Yes

No

No

Included

Abdominal

<72 h Yes Abdominal Included Unknown Included No No Yes No Yes Beinlich (1999)

179

a prospective clinical trial. The presence of verification bias could not be excluded in any of the studies. The number of patients included in each study varied between 73 and 833. Most studies included a diverse group of pregnant women at risk for preterm delivery. However, one study explicitly excluded women with diabetic pregnancies [23], one explicitly excluded women with premature rupture of the membranes [19], and two studies explicitly excluded women with multiple pregnancies [24,30]. The minimum gestational age varied between 22 and 33 weeks, whereas the maximum gestational age varied between 37 and 42 weeks. In four studies women were treated with corticosteroids [18,22,25,29]. For all studies, thin layer chromatography was used to determinetheL/Sratiointheamnioticfluid,whichwascentrifuged priortothedeterminationintwostudies[28,30].Fortheremaining two studies, it was unclear whether the specimen was centrifuged [18,29]. To assess the LBC, resistive-pulse counting of lamellar bodies with the platelet channel of a standard hematology cell counter was used in all studies. Four studies assessed the LBC in uncentrifugedamnioticfluidspecimen[20,22,24,27],whereasthe other eight studies used centrifuged amniotic fluid samples [19,21,23,25,26,28–30]. Seven studies described more than one cut-off value to indicate pulmonary maturity [18– 20,22,24,27,29].FortheL/Sratio,thecut-offvaluesvariedbetween 2.0 and 2.5, and for the LBC between 6000 and 79,000 lamellar bodies per micro liter.

Clinical, radiological Clinical, radiological and therapy Clinical, radiological and therapy Clinical, radiological and therapy Radiological and therapy Therapy Clinical, radiological and therapy Clinical, radiological and therapy Clinical, radiological and therapy Clinical, radiological Diagnosed by standard criteria Clinical, radiological and therapy Therapy Unknown Unknown No Yes Abdominal/vaginal/cesarean Abdominal/vaginal/cesarean Excluded Unknown Unknown Unknown Partly Yes Abd El Aal (2005) Bahasadri (2005)

Yes No

Yes Yes

No No

No No

Included Unknown

Exclusion blood/ meconium stained samples PPROM Multiple pregnancies Prospective Cohort study Study (year)

Table 1 Key characteristics of the included studies.

Verification bias

Blinding

Consecutive series

Diabetic pregnancies

Amniotic fluid sampling

Interval sampling – delivery

Definition RDS

A.E. Besnard et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 169 (2013) 177–183

The prevalence of RDS varied between 6% and 32%. For the L/ S ratio, the sensitivity varied between 62% and 100%, whereas the specificity varied between 64% and 89%. For the LBC, the sensitivity varied between 73% and 99%, whereas the specificity varied between 60% and 100%. For both tests, the prevalence of RDS and sensitivity and specificity for each individual study are summarized in Tables 3 and 4, respectively. Fig. 1 illustrates all reported cut-off values for the L/S ratio and LBC with their corresponding sensitivity and specificity in a scatter plot. To provide further information on the different cut-off values, a subgroup analysis was performed to evaluate the sensitivity and specificity within a specific range of cut-off values. For the LBC, at a range of cut-off values from 15,000/ mL to 25,000/mL, the sensitivity is 76% (95% CI 57– 88%) with a specificity of 90% (95% CI 70–97%). The sensitivity is 94% (95% CI 18–100%) and the specificity is 75% (95% CI 13–99%) when exploiting a range of cut-off values from 45,000/mL to 57,000/mL. Due to the small number of studies reporting on the L/S ratio in this meta-analysis, data were found insufficient to execute a subgroup analysis and directly compare the two most frequently used cut-off values of 2.0 and 2.5. Therefore, we combined the studies reporting on both LBC and L/S ratio in the same population with the studies in Wijnberger’s meta-analysis [7–12,28–30]. The results are shown in Table 5. Figs. 2 and 3 show all the cut-off values with their corresponding sensitivity and specificity as well as summary ROC curves for the L/S ratio and LBC respectively of this subgroup analysis.

180

Table 2 Study quality per study of the 13 included studies assessed with the QUADAS checklist. Bahasadri (2005)

Beinlich (1999)

Chapman (2004)

Ghidini (2005)

Haymond (2006)

Karcher (2005)

Khazardoost (2005)

Neerhof (2001)

Piazze 1 (1999)

Piazze 2 (2005)

Roiz Hernandez (2002)

Winn MacMillan (2005)

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

No

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Unclear

Yes

Unclear

Yes

Unclear

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes Yes

Yes Yes

Yes Yes

Yes Yes

Yes Yes

Yes Yes

Yes Yes

Yes Yes

Yes Yes

Yes Yes

Yes Yes

Yes Yes

Yes Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Unclear

Yes

Yes

Yes

Yes

Unclear

Yes

Unclear

Yes

Yes

Yes

No

No

Yes

Unclear

No

Yes

Unclear

Unclear

Yes

Yes

Unclear

Unclear

Unclear

Unclear

Unclear

Unclear

Unclear

Unclear

Unclear

Unclear

Unclear

Yes

Unclear

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

No

No

No

No

No

Yes

No

No

No

No

No

No

No

Yes No

Yes Unclear

Yes Unclear

Yes Yes

Yes Unclear

Yes No

No Unclear

Yes Yes

Yes Yes

No Unclear

Yes Unclear

No Unclear

Yes Unclear

A.E. Besnard et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 169 (2013) 177–183

Patients representative of practice Clear description selection criteria Reference standard likely to detect RDS Time between test and reference standard short enough Complete verification Consistent reference standard Index test and reference standard performed independently Clear description of index test Clear description of reference standard Results index tests interpreted independent of results reference standard Results reference standard interpreted independent of results index tests Clinical data same as practice Uninterpretable data reported Withdrawals explained Intervention between index test and reference standard

Abd El Aal (2005)

A.E. Besnard et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 169 (2013) 177–183

181

Table 3 Performance of the L/S ratio in the prediction of RDS. If a study reported on multiple cut-off values, only one is shown. Study (year)

Number of patients

Number of cut-off values

Cut-off value

Karcher (2005) Neerhof (2001) Piazze (1999) Winn-McMillan I (2005) Winn-McMillan II (2005)

201 833 92 109 96

1 1 1 2 1

2.5 N/A 2.5 2.0 2.0

RDS

No RDS

Prevalence of RDS

TP

FN

FP

TN

8 82 14 9 8

5 18 5 0 0

20 174 26 20 14

168 559 47 80 74

0.06 0.12 0.18 0.08 0.08

Prediction of RDS Sens

Spec

0.62 0.82 0.74 1.00 1.00

0.89 0.76 0.64 0.80 0.84

TP = true positive; FN = false negative; FP = false positive; TN = true negative; sens = sensitivity; spec = specificity.

Table 4 Performance of the LBC ratio in the prediction of RDS. If a study reported on multiple cut-off values, only one is shown. Study (year)

Number of patients

Number of cut-off values

Cut-off value

Abd El Aal (2005) Bahasadri (2005) Beinlich (1999) Chapman (2004) Ghidini (2005) Haymond (2006) Karcher (2005) Khazardoost (2005) Neerhof (2001) Piazze I (1999) Piazze II (2005) Roiz-Hernandez (2002)

73 104 21 88 102 184 219 80 833 92 178 264

4 2 1 6 1 2 1 1 2 1 1 3

18,000 45,000 30,000 25,000 37,000 50,000 30,000 50,000 N/A 20,000 22,000 57,000

RDS

No RDS

Prevalence of RDS

TP

FN

FP

TN

18 22.7 5 13 16 11 11 17 89 18 44 36

5 0.3 1 1 1 1 2 3 11 1 17 3

0 1.3 5 9 31 69 51 18 266 20 21 65

50 79.7 10 65 54 103 155 42 467 53 96 160

0.32 0.22 0.29 0.16 0.17 0.07 0.06 0.25 0.12 0.21 0.34 0.15

Prediction of RDS Sens

Spec

0.78 0.99 0.83 0.93 0.94 0.92 0.85 0.85 0.89 0.95 0.73 0.92

1.00 0.98 0.67 0.88 0.64 0.60 0.75 0.70 0.64 0.73 0.82 0.71

TP = true positive; FN = false negative; FP = false positive; TN = true negative; sens = sensitivity; spec = specificity.

4. Discussion This meta-analysis demonstrates that the LBC is a good diagnostic test, having an accuracy similar to the L/S ratio, and by comparing the ROC curves it perhaps performs slightly better.

Fig. 1. Receiver-operating characteristics (ROC) of studies comparing lamellar body count and L/S ratio in their capacity to predict the occurrence of respiratory distress syndrome. Summary ROC curves are also given.

The overall results of this meta-analysis are concurrent with the meta-analysis in 2001 [14]. Wijnberger et al. analyzed six studies [7–12], comparing the L/S ratio and the LBC, and similarly concluded that the LBC performs slightly better than the L/S ratio (p = 0.13). One of the advantages of the current meta-analysis is that it provides information on the different cut-off values. The sROC curves of Fig. 1 give a general outline on the overall accuracy combining the sensitivity and specificity points of different cut-off values, whereas the subgroup analysis calculates the accuracy of a certain cut-off value. The subgroup analysis for the LBC showed that a high cut-off value, range 45,000–57,000, correlates with a higher sensitivity, whereas a low cut-off value, range 15,000– 25,000, correlates with a high specificity. As a result of the wide range of cut-off values described in the different studies, we were unable to determine the cut-off value with the best accuracy, which can subsequently be recommended for the clinical practice. By obtaining more data around several specific cut-off values with a small range, the best cut-off for clinical application can eventually be derived. A limitation of this meta-analysis is the clinical heterogeneity, i.e. study population, clinical characteristics and cut-off values, as well as the statistical heterogeneity, i.e. variation in results, of the included studies. Since a small number of articles were found between 1999 and 2009, this analysis could not be limited to studies directly comparing the LBC and L/S ratio in the same population. Therefore, this meta-analysis was obliged to compare both tests in studies with different study designs, clinical characteristics and differences in reported cut-off values. By solely analyzing studies with identical cut-off values, the heterogeneity concerning the cut-off values might be reduced, but this analysis would cause loss of data points or may even exclude complete studies if they did not report for that cut-off value. The statistical heterogeneity is taken into account by using a random

182

A.E. Besnard et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 169 (2013) 177–183

Table 5 Sensitivity and specificity of the different cut-off values in studies comparing LBC and L/S ratio in the same population [7–12,28–30].

LBC

L/S ratio

Cut-off value

Sensitivity

95% CI

Specificity

95% CI

5000–10,000 20,000 30,000 50,000–55,000

0.59 0.82 0.95 0.99

0.37–0.78 0.65–0.92 0.75–0.99 0.92–1.00

0.97 0.88 0.82 0.65

0.93–0.99 0.73–0.95 0.66–0.91 0.60–0.70

0.73 0.86 0.96

0.60–0.83 0.65–0.95 0.79–0.99

0.92 0.80 0.84

0.82–0.97 0.68–0.89 0.77–0.89

1.8–2.0 2.2–2.6 2.7–3.0

Fig. 2. Receiver-operating characteristics (ROC) of studies comparing lamellar body count and L/S ratio in the same population per cut-off value of L/S ratio in their capacity to predict the occurrence of respiratory distress syndrome [7–12,28–30].

effects model that estimates the amount of between-study variability in both sensitivity and specificity. Nevertheless, the heterogeneity could have influenced the results comparing both tests on accuracy. To reduce the heterogeneity of the included studies, an individual patient data meta-analysis should be used in which the data of all the individual patients, the detailed patient characteristics and the associated results are explored. This way the possible interactions between patient factors and accuracy can be evaluated. Another problem in this meta-analysis is that 12 of the detected articles reported on the performance of the LBC versus four that reported on the L/S ratio. As stated before, the L/S ratio is currently the gold standard, but it has limitations. In the search for an alternative, the LBC is a promising option. Publication of only those papers that report positive or topical results leads to publication bias. If publication bias is present, the accuracy of the LBC reported in this meta-analysis was most likely overestimated. In all the mentioned studies, a random cut-off value was chosen irrespective of clinical parameters, such as gestational age. It is therefore possible that a certain value of LBC or L/S ratio indicates FLM at one gestational age, but immaturity at another. A solution to this problem lies in adjusting the cut-off value according to gestational age. Previous studies show that increased gestational age has a positive effect on the performance of both L/S ratio and LBC, but fewer studies have reported on this effect on LBC. It is also possible that the test performance is influenced by clinical factors,

Fig. 3. Receiver-operating characteristics (ROC) of studies comparing lamellar body count and L/S ratio in the same population per cut-off value of LBC in their capacity to predict the occurrence of respiratory distress syndrome [7–12,28–30].

such as fetal growth, the total amount of amniotic fluid, the presence of blood, meconium or infection or diabetic status of the mother. Future research should focus on these clinical factors and determine their effect on the performance of both tests, especially the effect of the amount of amniotic fluid on LBC performance [31– 33]. In clinical practice, it is important to predict the presence of RDS accurately in order to prevent infants being born with immature lungs and the associated complications. Consequently, the ideal diagnostic test for FLM should have a high sensitivity and a high negative (mature) predictive value. This also implies a high number of false positives. Delaying the delivery incorrectly, i.e. in the presence of FLM, can have negative consequences for child and mother, e.g. in cases of pre-eclampsia, but in the majority of cases the consequences of RDS are probably limited. In conclusion, this meta-analysis illustrates that the LBC is a good measure to predict the occurrence of RDS, with an equal, if not slightly better, performance compared to the L/S ratio. Since the LBC is easy to perform, rapid, inexpensive, and available to all hospitals 24 h per day, we suggest that it should replace the L/S ratio in the assessment of FLM. Conflict of interest The authors declare that they have no conflict of interest.

A.E. Besnard et al. / European Journal of Obstetrics & Gynecology and Reproductive Biology 169 (2013) 177–183

References [1] Breathing in America: diseases, progress, and hope. Schraufnagel DE, editor. Respiratory distress syndrome of the newborn. The American Thoracic Society; 2010. [2] Grenache DG, Gronowski AM. Fetal lung maturity. Clinical Biochemistry 2006;39:1–10. [3] Sweet D, Bevilacqua G, Carnielli V, et al. European consensus guidelines on the management of neonatal respiratory distress syndrome. Journal of Perinatal Medicine 2007;35:175–86. [4] American College of Obstetricians and Gynecologists. ACOG Practice Bulletin No. 97: fetal lung maturity. Obstetrics and Gynecology 2008;112:717–26. [5] Gluck L, Kulovich MV, Borer Jr RC, Brenner PH, Anderson GG, Spellacy WN. Diagnosis of the respiratory distress syndrome by amniocentesis. American Journal of Obstetrics and Gynecology 1971;109:440–5. [6] Dubin SB. Characterization of amniotic fluid lamellar bodies by resistive-pulse counting: relationship to measures of fetal lung maturity. Clinical Chemistry 1989;35:612–6. [7] Ashwood ER, Palmer SE, Taylor JS, Pingree SS. Lamellar body counts for rapid fetal lung maturity testing. Obstetrics and Gynecology 1993;81:619–24. [8] Bowie LJ, Shammo J, Dohnal JC, Farrell E, Vye MV. Lamellar body number density and the prediction of respiratory distress. Clinical Chemistry 1991;95:781–6. [9] Dalence CR, Bowie LJ, Dohnal JC, Farrell EE, Neerhof MG. Amniotic fluid lamellar body count: a rapid and reliable fetal maturity test. Obstetrics and Gynecology 1995;86:235–9. [10] Fakhoury G, Daikoku NH, Benser J, Dubin NH. Lamellar body concentrations and the prediction of fetal pulmonary maturity. American Journal of Obstetrics and Gynecology 1994;170:72–6. [11] Greenspoon JS, Rosen DJD, Roll K, Dubin SB. Evaluation of lamellar body number density as the initial assessment in a fetal lung maturity test cascade. Journal of Reproductive Medicine 1995;40:260–6. [12] Lee IS, Cho YK, Kim A, Min WK, Kim KS, Mok JE. Lamellar body count in amniotic fluid as a rapid screening test for fetal lung maturity. Journal of Perinatology 1996;16:176–80. [13] Lewis PS, Lauria MR, Dzieczkowski J, Utter GO, Dombrowski MP. Amniotic fluid lamellar body count: cost-effective screening for fetal lung maturity. Obstetrics and Gynecology 1999;93:387–91. [14] Wijnberger LD, Huisjes AJ, Voorbij HA, Franx A, Bruinse HW, Mol BW. The accuracy of lamellar body count and lecithin/sphingomyelin ratio in the prediction of neonatal respiratory distress syndrome: a meta-analysis. BJOG An International Journal of Obstetrics and Gynaecology 2001;108:583–8. [15] Lijmer GJ, Mol BW, Heisterkamp S, et al. Empirical evidence of design-related bias in studies of diagnostic tests. JAMA Journal of the American Medical Association 1999;282:1061–6. [16] Whiting P, Rutjes AW, Reitsma JB, et al. The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews. BMC Medical Research Methodology 2003;10:25. [17] Reitsma JB, Glas AS, Rutjes AW, Scholten RJ, Bossuyt PM, Zwinderman AH. Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews. Journal of Clinical Epidemiology 2005;58:982–90.

183

[18] Winn-McMillan T, Karon BS. Comparison of the TDx-FLM II and lecithin to sphingomyelin ratio assays in predicting fetal lung maturity. American Journal of Obstetrics and Gynecology 2005;193:778–82. [19] Abd El Aal DE, Elkhirshy AA, Atwa S, El-Kabsh MY. Lamellar body count as a predictor of neonatal lung maturity in high-risk pregnancies. International Journal of Gynaecology and Obstetrics 2005;89:19–25. [20] Bahasadri S, Changizi N. Association between lamellar body count and respiratory distress in neonates. Saudi Medical Journal 2005;26:1414–6. [21] Beinlich A, Fischass C, Kaufmann M, Schlosser R, Dericks-Tan JS. Lamellar body counts in amniotic fluid for prediction of fetal lung maturity. Archives of Gynecology and Obstetrics 1999;262:173–80. [22] Chapman JF, Ashwood ER, Feld R, Wu AH. Evaluation of two-dimensional cytometric lamellar body counts on the ADVIA 120 hematology system for estimation of fetal lung maturation. Clinica Chimica Acta 2004;340:85–92. [23] Ghidini A, Poggi SH, Spong CY, Goodwin KM, Vink J, Pezzullo JC. Role of lamellar body count for the prediction of neonatal respiratory distress syndrome in non-diabetic pregnant women. Archives of Gynecology and Obstetrics 2005;271:325–8. [24] Haymond S, Luzzi VI, Parvin CA, Gronowski AM. A direct comparison between lamellar body counts and fluorescent polarization methods for predicting respiratory distress syndrome. American Journal of Clinical Pathology 2006;126:894–9. [25] Khazardoost S, Yahyazadeh H, Borna S, Sohrabvand F, Yahyazadeh N, Amini E. Amniotic fluid lamellar body count and its sensitivity and specificity in evaluating of fetal lung maturity. Journal of Obstetrics and Gynaecology 2005;25:257–9. [26] Piazze JJ, Maranghi L, Cerekja A, et al. Amniotic fluid lamellar body counts for the determination of fetal lung maturity: an update. Journal of Perinatal Medicine 2005;33:156–60. [27] Roiz-Hernandez J, Navarro-Solis E, Carreon-Valdez E. Lamellar bodies as a diagnostic test of fetal lung maturity. International Journal of Gynaecology and Obstetrics 2002;77:217–21. [28] Karcher R, Sykes E, Batton D, et al. Gestational age-specific predicted risk of neonatal respiratory distress syndrome using lamellar body count and surfactant-to-albumin ratio in amniotic fluid. American Journal of Obstetrics and Gynecology 2005;193:1680–4. [29] Neerhof MG, Haney EI, Silver RK, Ashwood ER, Lee IS, Piazze JJ. Lamellar body counts compared with traditional phospholipid analysis as an assay for evaluating fetal lung maturity. Obstetrics and Gynecology 2001;97:305–9. [30] Piazze JJ, Anceschi MM, Maranghi L, Porpora MG, Cosmi EV. The biophysical/ biochemical test. A new marker of fetal lung maturity in borderline cases. Journal of Reproductive Medicine 1999;44:611–5. [31] Wijnberger LD, de KM, Voorbij HA, et al. The effect of clinical characteristics on the lecithin/sphingomyelin ratio and lamellar body count: a cross-sectional study. Journal of Maternal-Fetal and Neonatal Medicine 2003;14:373–82. [32] Hunink MGM, Richardson DK, Doubilet PM, Begg CB. Testing for fetal pulmonary maturity: ROC analysis involving covariates, verification bias, and combination testing. Medical Decision Making 1990;10:201–11. [33] St Clair, Norwitz ER, Woensdregt K, et al. The probability of neonatal respiratory distress syndrome as a function of gestational age and lecithin/sphingomyelin ratio. American Journal of Perinatology 2008;25:473–80.