False-Positive Results in Point-of-Care Ovulation Prediction Devices Due to Very Low Concentrations of Human Chorionic Gonadotropin

Point of care devices which detect luteinizing hormone (LH) are used to predict ovulation and time intercourse in women who are trying to get pregnant. Women attending fertility clinics also commonly use these devices to time intrauterine insemination.  Although the hormones LH and hCG share 80% structural homology, cross reactivity in quantitative (laboratory) LH and hCG assays has not been a problem for many years due to the use of very specific antibody pairs. Many physicians and laboratorians assume that that specificity holds true for qualitative (home) devices as well.

Recently, a women undergoing fertility treatment at our institution detected a positive LH surge using an over the counter LH device despite the fact that she was later found to be pregnant. This made us ask the question "Could the over-the-counter ovulation kits cross react with hCG?"  Therefore, we undertook a study where we added purified hCG to saline and tested three home ovulation prediction devices [Clear Blue® (Swiss Precision Diagnostics, Geneva, Switzerland), First Response® (Church & Dwight, Princeton, NJ), and Walgreens® (Inverness Medical (now Alere), Waltham, MA)]. We found that all the devices we tested returned false positive results at hCG concentrations ranging from 10 to 10,000 mIU/mL.

The concentration of hCG at which devices were positive varied by brand. Both Clear Blue and Walgreens were clearly positive at hCG concentrations of 100 mIU/mL! Clear Blue was positive at an hCG concentration of 10 IU/L! Walgreens turned positive between 50 and 100 mIU/mL, and First Response turned positive between 5,000 and 10,000 mIU/mL. The LH concentration that caused positive results also varied by device brand (Table). Only Clear Blue was definitively positive at 50 mIU/mL of LH. First Response and Walgreens turned positive between 50 and 100 mIU/mL of LH.

These devices may produce false positive results in women who are very early in pregnancy. At our institution, the reference interval for hCG in the first three to four weeks of pregnancy is 9 to 130 mIU/mL. Package inserts for some devices contain a cautionary statement that results obtained during pregnancy or administration of certain drugs including hCG may produce misleading results; however, it is clear that most physicians and laboratorians are unaware of the any potential cross-reactivity. This data is being presented at the American Association of Clinical Chemistry (AACC) meeting (Wednesday July 31, 2013, Poster B50) in Houston, and will be published in full later this year in Clinical Biochemistry doi.org/10.1016/j.clinbiochem.2013.07.017.

Fertility clinics and physicians that rely on home LH devices to detect an LH surge for the timing of intrauterine insemination should be aware that early pregnancy may cause false positive results on LH devices. Fertility clinics in particular should instruct their patients to use home LH devices with minimal hCG cross-reactivity.

Predicting the Success of In Vitro Fertilization

IVFWhat are my chances of getting pregnant? More specifically, if I am having trouble getting pregnant, will I be successful if I undergo the process of in vitro fertilization (IVF)? These are important questions especially since infertility treatments (especially IVF) is time consuming and costly.  

There are tools to assess a woman's fertility status. For instance, Fertistat  is a free on-line questionnaire that was developed at Cardiff University in the UK. Its primary function is as a fertility awareness tool.

A recent publication by Choi et al  reported a personalized model for predicting the success rate of women undergoing their first cycle of IVF. It should be noted that this method is now patented and marketed by Univfy Inc. and this blog post is in not an endorsement for this product.  Their model utilizes patient demographics and reproductive history, age at time of first IVF, BMI, smoking status, gravity, parity, pregnancy losses before 20 weeks, number of ectopic pregnancies, antral follicle count, day 3 serum FSH the year, patient diagnosis, male partner reproductive health including age, total motile sperm count, use of sperm extraction method and use of donor sperm. They developed their model using data from 13,076 first IVF treatment cycles performed at three clinics in Spain, Canada, and the United States. Then they created their PreIVF-D by combining the data from the three clinics and they used a training set of 1,061 independent cases. Finally they tested their model using another set of data from 1,058 patients.  They found that the most important prognostic contributors to their model were patient age, total motile sperm count, BMI, day 3 serum FSH, and antral follicle count.

The authors conclude that age-based estimates of live birth probabilities in IVF treatment are not optimal and that their PreIVF-D model performs better with a 35.7% improvement in the ability to predict live birth. Notably, the area under the curve (AUC) for the age-based prediction was 0.614 and for Pre-IVF-D it was 0.634.  This means that the two models were essentially equivalent.  There are two limitations to this study that should be pointed out. First, the authors discuss that it is not possible for them to prove that PreIVF-D works for a clinic outside this study. In fact, there may be clinic-specific trends (such as referral types, regional BMI, and treatment approaches) that significantly affect their predictive model. Second, the authors do not mention the methods that each clinic used for FSH measurement. According to 2013 CAP surveys, there are up to 2 fold differences in FSH results depending on the method used. It would be interesting to know if the clinics utilized one FSH method or several as this may contribute to significant clinic-to clinic variations.

Personalized prognostic tools are likely the way of the future in reproductive medicine & infertility treatment. Although their utility now may be limited, they will undoubtedly continue to improve and evolve just as predictive models for assessing risk of downs syndrome has evolved over the past 20 years.   

TMI? A home hCG test that detects pregnancy and estimates weeks since conception

CalendarI'm beginning to wonder what could possibly come next. Last month, Swiss Precision Diagnostics (via its Procter & Gamble partner) unveiled its newest product in consumer diagnostics: the Clearblue Advanced Pregnancy Test with Weeks Estimator. This urine hCG test determines pregnancy status but also provides an estimate of the number of weeks since ovulation. The device has been cleared by the FDA and will be availabe in the US in September, 2013.

It works like other qualitative urine hCG tests but the body of the device contains two test strips that capture and detect the hormone in the urine sample. Detection of hCG is accomplished by the appearance of a colored band at a specific location on the test strip.

One test strip is designated "high sensitivity" and can detect a low concentration of hCG (detection limit of 10 IU/L). This test strip determines pregnancy status (pregnant or not pregnant). The other test strip has "low sensitivity" and detects higher concentrations of hCG and is used to estimate the number of weeks since conception. An optical reader housed within the device determines the color intensity of the test strips and a digital display reports the results as 1) Not pregnant; 2) Pregnant 1-2 weeks; 3) Pregnant 2-3 weeks; or 4) Pregnant 3+ weeks. The device reports an invalid result if there is a malfunction due to the device itself or operator error.

Clearblue with Estimator

Note that the weeks estimate is based on hCG concentration which is not how a pregnancy is usually dated. Physicians calculate gestational age by the day of the last menstrual period (LMP) so the device's estimate will be about 2 weeks less than one based on the LMP.

In its press release, Procter & Gamble states that the test is "more than 99 percent accurate in detecting pregnancy from the day of the expected period, and it is approximately 93 percent accurate in estimating how many weeks based on time since ovulation." I wanted to know what studies were done to support those claims but I received no response when I reached out to the individual at Procter & Gamble identified in the press release. However, there are some data included in the FDA's decision summary:

  • An early pregnancy study was conducted using 100 urine samples collected from non-pregnant women expecting to become pregnant. These samples were collected on days -6 to 1+ relative to the day of expected period. 99.0% of the devices gave a "pregnant" result by day zero (the day of the expected period).
  • A clinical study was conducted using samples from 153 volunteers with singleton pregnancies to evaluate performance of the “Weeks Estimator” feature compared to actual gestational age (method not identified). Agreement of “Weeks Estimator” with actual gestational age ranged from 45-99% (bold-faced emphasis is mine).

The bolded statement above is ambiguous but it likely is supposed to mean that the device is accurately able to estimate the true week of gestational age 45-99% of the time. If so then there are several questions that need to be answered. For example, what was the source of the range? Was it derived from different studies or from different gestational ages? What was the median gestational age? Is the device more accurate at certain gestational agess? Also, the range doesn't indicate how inaccurate the device can actually be. That is, when it is wrong how wrong is it? 1 week, 2 weeks, 3 weeks, or even more?

Given that the "Weeks Estimator" is highly variable, it is likely that Procter & Gamble will include language (similar to what the FDA stated) to caution the consumer that "the 'Weeks Estimator' is meant solely as an estimate for the consumer and is not intended as a substitute for a doctor’s clinical diagnosis. The ‘Weeks Estimator’ is not intended for multiple pregnancies. The estimate provided by the device may be inaccurate in these cases."

From a practical perspective, while women may want/need to know when they conceived, this is not the device to accomplish the job. It seems to provide highly variable information that, at best, is just a curiosity. Procter & Gamble says “confirming pregnancy is a life-changing moment in any woman’s life, and it sparks so many immediate questions like 'when did I get pregnant?'" True, but necessary? I'm not convinced.

NIH Consensus Meeting on Diagnosis of GDM

Diabetes definitionDavid has blogged in the past about the diagnosis of gestational diabetes mellitus (GDM). In July 2012 he discussed a debate that was underway among experts regarding newly proposed diagnostic guidelines.

Just to recapitulate the debate, for at least 10 years we have been diagnosing GDM with a two-step process.

  1. First, there is a screening test performed by giving a non-fasting woman a 50-gram dose of glucose and then measuring her serum glucose 1 hr later. If the woman's glucose concentrations were higher than expected in that screen, then she went on to a diagnostic test.
  2. For the diagnostic test, a fasting patient is given 100-gram of glucose and then serum glucose concentrations are measured at 0, 1, 2 and 3 hours.

In 2010, the International Association of Diabetes in Pregnancy Study Groups (IADPSG) made recommendations for glucose tolerance testing in pregnancy based on the results of a study called HAPO (Hyperglycemia and Adverse Pregnancy Outcomes).  That study clearly demonstrated that the risks of adverse maternal and fetal outcomes continually increase as maternal glucose concentrations increase. In 2011, the American Diabetes Association adopted these new diagnostic criteria. In the new diagnostic approach, a 75-gram load is given to fasting women and blood is collected at 1 and 2 hours. However, also in 2011, the American College of Obstetricians and Gynecologists issued a statement indicating:

"Diagnosis of GDM based on the one-step screening and diagnosis test outlined in the International Association of Diabetes in Pregnancy Study Group guidelines is not recommended at this time because there is no evidence that diagnosis using these criteria leads to clinically significant improvements in maternal or newborn outcomes and it would lead to a significant increase in health care costs."

This is not surprising coming from ACOG. Their position is always to ask for evidence that new protocols: a) positively affect outcomes; and, b) do not harm the mother or infant. However, this division between the ADA and ACOG left everyone with the debate that David discussed in July 2012.

In March 2013, the National Institutes of Health (NIH) held a consensus development conference which convened an independent panel of health professionals and public representatives. During the conference invited experts presented and discussed current scientific data.

They addressed the following questions:

    1. What are the current approaches for GDM, what are the glycemic thresholds for each approach, and how were they chosen?
    2. What are the effects of various screening/diagnostic approaches for patients, providers, and U.S. healthcare systems?
    3. In the absence of treatment, how do the outcomes of mothers and their offspring compare with those who do not?
    4. Does treatment modify the health outcomes of mothers and their offspring?
    5. What are the harms of treating?
    6. Given all of the above, what diagnostic approach(es) for gestational diabetes mellitus should be recommended, if any?
    7. What are the key research gaps in the diagnostic approach of gestational diabetes mellitus?

      The committee felt that a one-step approach would, in many ways, be advantageous over the two-step approach. First, the current two-step approach is not used other than during pregnancy and is largely restricted to the United States. Second, there would be value in a consistent diagnostic approach across an individual's lifespan, within the United States, and during pregnancy around the world. This would allow better standardization of best practices and comparability of research outcomes. The one-step approach would also allow a diagnosis to be made within a single healthcare visit.

      The committee felt that there is good evidence that increasing glucose concentrations during pregnancy are associated with greater maternal and perinatal morbidities. They also state that there is evidence that treatment of women with GDM—diagnosed either by the one-step or two-step approach—may improve some outcomes. However, the new one-step approach, as proposed by the IADPSG, is anticipated to increase the diagnosis of GDM by 2-3 fold, to a prevalence of approximately 15-20%. Because it is unclear if these women will benefit from treatment and these additional diagnoses will increase health care costs, the consensus committee concluded that the old two-step method for diagnosis should be retained.

      "…at present, the panel believes that there is not sufficient evidence to adopt a one-step approach, such as that proposed by the IADPSG. The panel is particularly concerned about the adoption of new criteria that would increase the prevalence of GDM, and the corresponding costs and interventions, without clear demonstration of improvements in the most clinically important health and patient-centered outcomes. Thus, the panel recommends that the two-step approach be continued. However, given the potential benefits of a one-step approach, resolution of the uncertainties associated with its use would warrant reconsideration of this conclusion."

      The panel went on to identify 9 areas of research that are needed including outcomes and cost benefit ratio studies. This certainly lays the groundwork for years of future studies. In the meantime, there seems to be a great deal of support for maintaining the two-step approach for diagnosis of GDM.

      More on noninvasive prenatal testing for fetal aneuploidy

      We have written about nonivasive prenatal testing (NIPT) on this blog several times.  Because they are so new, the landscape around these tests is continually evolving.  The American College of Obstetricians and Gynecologists (ACOG) published guidelines on these tests in December of last year.  Just this week, the American College of Medical Genetics and Genomics (ACMG) released its policy statement on the same topic.  Note that the ACMG refers to these tests as "noninvasive prenatal screening" (NIPS) tests to emphasize that this is what they are: screening, not diagnostic tests.

      The ACMG calls for caution before these tests become widely integrated into prenatal care due to the current lack of data obtained from prospective clinical trials.  While they acknowledge that NIPS tests have high sensitivity and specificity there are limitations to the technology and false-positive and false-negative results do occur.

      A particular concern, and one that doesn't get as much attention as it should, is that most of the fetal DNA in the mother's blood sample originates from the placenta and not the fetus and it may not accurately reflect the fetal karyotype.  They emphasize (as have others), that abnormal NIPS test results must be confirmed by invasive diagnostic tests such as amniocentesis.

      The policy statement also lists several limitations to NIPS tests.  Among them:

      • They only detect aneuploidies (and some detect sex chromosome abnormalities).
      • Certain chromosome abnormalities are not detected.
      • The tests take longer to perform and result than more well-established tests.
      • Data on the performance of the tests in twin and triplet pregnancies is not well established.

      A recent paper published in the journal Obstetrics and Gynecology has a similar ring to it.  The authors make several interesting observations:

      • First, they point out that well-established tests were developed in academic settings and came into use gradually and only after independent clinical studies generated data to support their use.  In contrast, NIPT (also developed in academic settings) was quickly licensed to commercial enterprises that have brought them to market without FDA review (as these are "lab-developed tests," FDA appoval is not required).
      • From the analytical perspective, there are currently no guidelines regarding quality control and quality assurance for NIPT; a vital component of any lab test.
      • The performance of NIPT in actual clinical practice settings (i.e. not a clinical study) is currently not well known or documented.  This is especially true for populations of women that have not been represented in the clinical studies (e.g. woman at low risk for having a fetus with an aneuploidy).
      • The more well-established tests are able to detect fetal anomalies besides aneuploidy (e.g. open neural tube defects).

      The authors also reflect on how NIPT should be incorporated into clinical care.  They agree with the ACOG recommendations that the tests should not be offered to low-risk women but they go a bit further and state that the most appropriate use of NIPT is as a second screening test used for those who have an abnormal result from convential, more well-established screening tests.  The latter point is something I have commented on before and I could not be in more complete agreement.

        Vitamin D During Pregnancy

        HiResIn the last several years, there have been a lot of articles in both the popular press and the scientific literature about Vitamin D. There are studies that report that low concentrations of vitamin D are associated with everything from cancer to multiple sclerosis and asthma to  cardiovascular disease. There has been a lot of debate about what to conclude from these studies. Whole books could be written about this, but just to summarize, some of the controversial issues include:



        1. What concentration of vitamin D in the blood should we consider normal?
        2. What is the best method to measure vitamin D?
        3. If we increase our concentrations of vitamin D will the risk for things like cancer, multiple sclerosis, asthma, and cardiovascular disease decrease?
        4. How much vitamin D should people take daily, if any?

        What about vitamin D during pregnancy? There have been many, many, studies examining the role of vitamin D in pregnancy.  Observational studies have suggested that lower vitamin D intake is associated with preeclampsia and gestational diabetes or higher blood glucose, as well as adverse fertility parameters and bacterial vaginosis. These are covered in a 2012 systematic review.  Vitamin D status has also been examined in recurrent pre-term birth but in a recent 2012 study, the authors concluded that vitamin D status at mid-pregnancy was not associated with recurrent pre-term birth. 

        Two studies have already come out this year which really only confuse the picture.  In January 2013 Gernand, et al published a study in which they had measured vitamin D in 2,146 serum samples that had been collected from 1959 to 1965. The serum was collected at gestational age of 26 weeks or less from women who delivered singleton, live birth infants.  They reported that higher vitamin D concentrations correlated with higher infant birth weight and larger infant head circumference.  They conclude that randomized controlled trials are needed to test if maternal vitamin D supplementation can improve fetal growth.

        This study might make you think supplementation during pregnancy should be started right away. However, in another study published in January 2013, Weisse, et al published a paper  that examined vitamin D concentrations in 378 mothers (collected at 34 weeks gestation) and cord blood (collected at delivery).  They found that the vitamin D in maternal and cord blood was highly correlated with each other and both showed a seasonal variation. They also reported that higher vitamin D concentrations in maternal blood (and cord blood) were positively associated with the child's risk of food allergies within the first 2 years of life. In other words higher vitamin D was associated with higher risk for allergy. They conclude that their study argues against vitamin D supplementation during pregnancy to protect the infant against allergy.

        In summary, the studies on vitamin D in pregnant women are not much different than those in non-pregnant women: That is, there are many associative studies, but further interventional outcome based studies are needed. In the mean time, pregnant women should eat a balanced diet and take the prenatal vitamins their physicians recommend.

        Laboratory testing for premature rupture of membranes

        This post is by a guest author, Douglas Stickle, Ph.D.  Dr. Stickle is a professor in the Department of Pathology at Thomas Jefferson University and the director of chemistry and point-of-care testing at Jefferson University Hospitals in Philadelphia, PA.

        Rupture of membranes (ROM) is the term used to describe the breaking of the amniotic sac, as normally occurs before the onset of labor. If this happens earlier than the 37th week of pregnancy it is called preterm ROM (PROM). It’s a condition that can lead to a preterm birth, or, if very early, a preterm, premature birth.

        Preterm baby 2When PROM happens, there is an increased risk of complications due to intrauterine infection, umbilical cord compression, and the neurodevelopmental disorders that are associated with a preterm delivery. Diagnosis of PROM is particularly important when the gestational age is incompatible with a viable birth, often considered to be a fetal age less than 24 weeks. In such cases, medical intervention is necessary to preserve the chances for a live birth.

        Suspected cases of PROM are often investigated by laboratory analysis of fluid obtained from the vagina to detect properties or substances that should otherwise not be present unless the fluid contains amniotic fluid due to PROM. The simplest forms of testing are measurement of acidity (pH) of the fluid, or a test called "fern" testing. Fern testing refers to the fern-like appearance of amniotic fluid when it is dried on a glass slide. Both of these tests aren’t very accurate and so other tests have been developed to better identify patients with ruptured membranes.

        These other tests are designed to detect molecules that are normally present in amniotic fluid but not vaginal fluid. For example, tests have been developed that detect alpha-fetoprotein (AFP) or insulin-like growth factor binding protein-1 (IGFBP-1). The presence or absence of these molecules in the specimen are determined by a lateral flow immunoassay. The assay works like commonly performed tests for human chorionic gonadotropin (hCG) (aka pregnancy tests).

        These tests are highly sensitive to low concentrations of these molecules, which is both good and bad. It’s good because they can detect small amounts of the molecules and lead to a more accurate diagnosis. It’s bad because these two molecules are also present in maternal blood which means that if a sample is contaminated with blood, the certainty of a positive test to detect amniotic fluid is called into question.

        From the doctor’s perspective, a practical advantage of the immunoassays is that their results are binary – the result is either positive or negative — whereas the pH test and the fern test are more subjective and difficult to interpret definitively. However, the AFP and IGFBP-1 tests may be subject to false-positive results as the gestational age of the fetus approaches term. This suggests that, at later stages of pregnancy, these biomarkers may signify imminence of delivery.

        The gold standard, or best test, to diagnose rupture of membranes is a dye test, in which a colored fluid is injected into the amniotic fluid followed by direct observation to see if the dye subsequently appears in the vaginal pool fluid. Also, ultrasound imaging of the amniotic fluid volume may also assist in diagnosis of PROM, but in individual cases such imaging may be difficult to interpret. Given the low but finite risk of complications of the dye test, the AFP and IGFBP-1 tests are often preferred as first-line tests for preterm premature rupture of membranes.

        Should DNA-based tests for Down syndrome screening replace biochemical tests?

        In a previous post I described the clinical performance of DNA-based screening tests for fetal aneuploidies like Down syndrome.  Overall, these tests have excellent detection rates (~99%) with very low false-positive rates (~0.2%).  In other words, these tests are about 99.0% sensitive and 99.8% specific.

        With performance like that one might expect these to be considered diagnostic tests.  They are not! Although quite good, test results must not be interpreted as definitive evidence that a fetus does or does not have an aneuploidy.  Recent recommendations from the American College of Obstetricians and Gynecologists (ACOG) are quite clear on that issue.

        In those same recommendations, ACOG also states that DNA-based screening tests may be performed only on women who are at increased risk of having a fetus with aneuloidy.  Among the indications listed for women considered to be at increase risk are:

        • Maternal age 35 years or older at delivery
        • Fetal ultrasound findings suggesting aneuploidy
        • A previous aneuploid pregnancy
        • Abnormal biochemical screening test results
        The ACOG is right to avoid recommending that DNA-based screening tests are acceptable to use regardless of risk factors.  Unfortunately, many women who are not at increased risk are using these new tests as a primary screening test and that's not a good idea.

        To understand why, considered a population of 100,000 pregnant women from the general population and assume that the prevalence of Down syndrome is 1 in 500 pregnancies.  That means that there would be 99,800 unaffected pregnancies and 200 pregnancies with Down syndrome.  The table below compares the results of the most commonly used biochemical screening test (the Quad test) to a DNA-based screening test.

        Quad vs DNA performance
        Clearly, the DNA-based test has several advantages over the Quad test.  Its positive predictive value is nearly 17 times greater than the Quad's and a positive DNA-based test result substantially increases the odds of having an affected fetus.  So why not use the DNA-based test as a primary screening test?  For the following reasons:
        • No studies have been published that have evaluated the performance of DNA-based tests in women who are not at increased risk of having a fetus with an aneuploidy
        • DNA-based tests are not widely available
        • The time it takes to report results of DNA-based testing is about 3 times greater than it is with biochemical testing
        • DNA-based tests are considerably more expensive than biochemical tests
        • Relative lack of insurance coverage for DNA-based tests
        Until these these limitations can be resolved, it makes more sense to use DNA-based testing as a secondary screening test.  In other words, it is only done after one of the risk factors described by ACOG (above) are met.  Doing so greatly improves the performance of both tests (see figure below).  A limitation of this approach is that the detection rate is that of the biochemical test which is not as high as it is with the DNA-based test.  Still, given the current limitations of DNA-based testing, this 2-step testing approach makes the most sense.
        DNA as secondary test

        The clinical utility of fetal lung maturity testing revisited 20 years later

        There is a saying in science that every 10 or 20 years scientists "reinvent" things. This refers to observations someone made and published, but the findings were largely ignored for 10-20 years until a new scientist comes along and makes the same or similar observation and suddenly everyone takes notice.  It seems to me that is what is happening with fetal lung maturity testing.

        As early as 1993 Wigton et al made the observation that in spite of documented fetal lung maturity (by L/S ratio or PG) major neonatal morbidity was observed in a population of 213 patients <37 weeks gestational age.  In 1997, Ghidini et al made a similar observation that the incidence of major neonatal complications among 153 preterm infants was high even in the presence of mature fetal lung tests.  I guess we didn't really pay attention to those papers since the Wigton paper has only been cited 16 times in 20 years and the Ghidini paper was cited 13 times in 16 years.

        In March of 2011 David discussed a study by Bates et al, that had been published in 2010 (17 years after Wigton's paper)  which showed the very same thing in a larger population. They demonstrated that even after documented fetal lung maturity (L/S ratio or PG), infants born 36 to 39 weeks (n=459) were at higher risk of adverse outcomes than infants born at 39 to 40 weeks (n=13,339). Infants born before 39 weeks were, overall, at 1.6-fold greater risk of having something bad happen to them.  Things like elevated serum bilirubin, ventilator support, low blood glucose, admission to a neonatal intensive care unit, or even RDS. People are beginning to take notice because this paper has already been cited 20 times in just 2 years!

        The Bates paper was followed in 2011 by a paper by Kamath et al Their study was not as large as Bates, but it broke down deliveries into late preterm (34 to 36 6/7 weeks; n=76) and early term (37 to 39 6/7 weeks; n=76) with documented fetal lung maturity as compared to infants greater than or equal to 39 weeks of gestation (n=262). These authors measured fetal lung maturity by TDx-FLM II, LBC, or PG. They again concluded that fetal lung maturity is insufficient to determine an infant's readiness for postnatal life.

        So here we are in 2013, twenty years after Wigton's paper, and the largest study of preterm infants with mature lung indices has just been published by Fang et al. This study was very similar to the Bates study and compares infants born 36 to 38 6/7 weeks gestation with mature fetal lungs (as determined by LBC, L/S, or PG; n=1011) to infants 39-41 weeks of gestation (n=11,701).  They found that delivery prior to 39 weeks with documented fetal lung maturity was associated with an 8.4% composite neonatal morbidity rate as compared to 3.3% for deliveries at 39 weeks or greater. This is compared to 6.1% and 2.5% respectively for the Bates study. Fang observed that a large proportion (49%) of women in their study who were undergoing an amniocentesis to determine fetal lung maturity, between 36 and 38 weeks of gestation, had pregestational or gestational diabetes. Because this could be a confounding factor in their results, they excluded all diabetics and reanalyzed their data. They found that even in non-diabetic patients, significantly higher rates of neonatal morbidity persisted in the group that was delivered <39 weeks.

        So what can we take away from these three "reinvented" papers? Certainly some pregnancy conditions require premature delivery. In these cases fetal lung maturity testing is irrelevant because the condition requires delivery regardless of lung maturity. In cases where premature delivery is not imminent, these studies show that gestational age itself has the strongest inverse correlation with morbidity.  Although fetal lung maturity testing may help to predict the absence of RDS, it does not mean that the infant will not have other complications due to immaturity. Essentially, delivery <39 weeks should be avoided regardless of fetal lung maturity testing. If lung maturity testing is performed, women should be counseled regarding the risk of neonatal morbidity even in the presence of a test results that indicates fetal lung maturity.

        As David concluded in 2011, "perhaps it is time to send these tests away once and for all".

        DNA-based tests for Down syndrome screening show excellent clinical performance

        The use of biochemical screening tests to identify pregnant women who are at high risk of having a fetus with Down syndrome is well established.  Biochemical screening began nearly 30 years ago and, over the years, the tests have evolved and improved.  Now there’s a new kid on the screening test block and it’s name is DNA.

        The discovery of cell-free fetal DNA in maternal plasma in 1997 opened up new possibilities for Down syndrome and other aneuploidy screening protocols.  Rather than rely on biochemical testing to determine a biochemical phenotype, DNA-based tests have been developed that can detect the molecular pathology of aneuploidies (e.g. a fetus that has more than the expected 2 copies of chromosomes 21, 18, or 13; the cause of Down syndrome, Edwards syndrome, and Patau syndrome, respectively).

        We’ve written about DNA-based screening tests before (here and here) and described the clinical performance of the Sequenom test.  Now, other clinical performance studies have been published for 3 of the 4 tests that are (or will be) commercially available.  As expected, all of them show excellent clinical performance.  As shown in the table below, the detection rates for trisomy 21 are greater than or equal to 99% with very low false-positive results.  Similar performance has been reported for trisomy 18 and 13.

        DNA test performance

        Table References: Genet Med 2011;13:913-920Genet Med 2012;14:296-305Obstet Gynecol 2012;119:890-901

        By comparison, the detection rate of the best biochemical Down syndrome screening test (the Integrated test) is very good at 93%.  However, about 5% of all Integrated test results are false-positive.  A 5% false-positive rate may not seem very high but it is.  For example, consider a population of 100,000 pregnant women who choose Integrated testing in the second trimester.  The prevalence of Down syndrome in the second trimester is about 1 in 500 pregnancies so 200 of those 100,000 women will have a fetus with Down syndrome and 99,800 women (100,000 – 200) will have unaffected fetuses.  Of those 99,800 women with unaffected fetuses, 4,900 will have a false-positive Integrated test result.

        Because the false-positive rate of the DNA-based tests is so low (about <0.2%), then if those same 100,000 women were screened there would be only 200 false-positive results, a 96% decrease!

        Does this mean that DNA-based tests should replace biochemical screening tests?  Probably not but I’ll leave the explanation as to why for my next post.