Category Archives: Meta-analysis

Meta-analysis

A New FDA-Approved Test for Predicting Preterm Delivery


Preterm baby

According to the March of Dimes, preterm birth occurs in approximately 10% of U.S. pregnancies. Until recently, cervicovaginal fetal fibronectin (fFN) was the only FDA-approved test for predicting preterm delivery in symptomatic women. We have blogged about fFN previously

Despite its FDA approval, fFN has limited clinical value. A condition with low prevalence, such as preterm delivery, has a low pre-test probability of occurring, hence a negative test result adds little to the assessment of the patient. Thus, a screening test for a low prevalence condition must demonstrate high positive predictive value (PPV) to be useful. The negative predictive value (NPV) of fFN is 99.5%, meaning a negative result is highly predictive that a woman will NOT deliver soon. However, PPV of fFN is only ~17%, meaning that less than 1 in 5 women with a positive test result will proceed to delivery with 7-14 days. For comparison, the PPV of flipping a coin in this population is 4%. Meta-analyses have supported the lack of utility for fFN.

In April 2018, the FDA approved cervicovaginal placental alpha macroglobulin-1 (PAMG-1), (brand name Parto Sure from QIAGEN) as a test for assessing the risk of spontaneous preterm birth in patients with symptoms of preterm labor.

Several recent studies have evaluated PAMG-1 for its ability to predict preterm birth.

Wing, et al. conducted a prospective study of pregnant women from 15 US sites, with signs or symptoms of preterm labor between 24 and 35 weeks of gestation with intact membranes and cervical dilations less than 3cm (>3 cm generally indicates active labor). They compared the utility of PAMG-1 to fFN. A summary of their key findings are shown in the table below.

Spontaneous preterm delivery ≤ 7 days

PPV

NPV

PAMG-1

19.0%

99.1%

fFN

6.5%

99.7%

     

Spontaneous preterm delivery ≤ 14 days

   

PAMG-1

25.0%

97.7%

fFN

11.1%

98.7%

Cervicovaginal PAMG-1 demonstrated similar negative predictive value and improved positive predictive value compared to cervicovaginal fFN.

Similarly, Melchor, et al. conducted a retrospective study of women with preterm contractions presenting to a single maternity hospital in Spain. They compared a one year period during which fFN was used to assess risk of pre-term delivery and a one year period where PAMG-1 was used. Similar to the Melchor study, patients were between 24-34 weeks of gestation with signs or symptoms of preterm labor and had intact membranes and a cervical dilation less than 3cm. A summary of their key findings are shown in the table below.

Spontaneous preterm delivery ≤ 7 days

PPV

NPV

PAMG-1

35.3%

98.3%

fFN

7.9%

97.9%

Both studies show improved positive predictive values for PAMG-1 over fFN. However, both studies reported sensitivities for PAMG-1 of 50%.  While this test can certainly be viewed as an improvement over fFN, PAMG-1 will only identify half of the women who will deliver within 7 day. Clearly a better marker to predict pre-term delivery is still needed.

Screening tests for group B strep infection


StreptococcusThe most common cause of life-threatening infections in newborns comes from a bacteria known as Streptococcus agalactiae (more commonly referred to as group B streptococcus or GBS).  This was stressed in a recent meta-analysis that reported that GBS infection remains an important, global cause of infant mortality.

The overall infection rate was 0.53 per 1,000 live births and, on average, about 10% of infected infants died.  Infants born in Africa were more likely to be infected (1.21 per 1,000) and die (22%) from the infection than infants born in the Americas or Europe (0.67-0.57 per 1,000 with 11 and 7% fatality rates).

It doesn't have to be this way because GBS infection is treatable with antibiotic therapy.  Indeed, in more developed countries, therapy is provided to women who carry the bacteria which prevents their baby becoming infected during delivery.  However, in poorer countries this is less likely to happen due to fewer resources.

Providing therapy to every pregnant women is not practical because not all women are colonized with GBS and so a key preventative strategy is to identify those women who do carry the bacteria.  The most sensitive test is culture performed on samples collected from the vagina and rectum.  The Centers for Disease Control and Prevention (CDC) published guidelines in 2010 that called for the routine GBS screening in all pregnant women at 35 to 37 weeks of gestation.  Testing needs to happen close to delivery (normally at ~40 weeks) because women can be colonized with GBS at anytime.  That is, a negative test result obtained earlier in pregnancy wouldn't rule-out the possibility that colonization then occured sometime after testing.  Women with a positive culture are treated with antibiotics during labor to prevent the transmission of GBS to their infant.

Although culture is considered the gold standard test for GBS screening, it is not perfect because some infants born to culture-negative women still get infected with GBS.  Also, culture techniques give results in 1–3 days, a time frame that may not be useful should an expectant mother go into labor prior to having the culture test performed.  For these women, DNA-based tests can be used.

These tests detect the presence of GBS using a DNA amplification technique like PCR and give results in a few hours rather than days.  Currently, these types of tests are not as sensitive as culture (i.e. they can give false-negative results) and so they aren't recommended for routine screening of women who are not in labor.  Their sensitivity is improved by using an enriched sample (one where the bacteria are allowed some time to multiply in a growth media), the use of this type of sample is impractical for women in labor when results are needed quickly.

Until an effective vaccine to prevent GBS infection is available, laboratory testing will remain an essential tool for identifying and preventing GBS.

Are there any good markers to predict preeclampsia?


David has written about preeclampsia in the past, but I thought I'd talk about some specific studies that have been published on that topic.

Recall that preeclampsia is when a pregnant woman develops high blood pressure and protein in the High blood pressure urine after the 20th week of pregnancy and it is usually associated with edema (swelling). Although preeclampsia occurs in only 5 to 8% of pregnancies, it is a major contributor of premature deliveries and neonatal morbidity in the United States. Because the etiology of preeclampsia is not well understood, the ability to predict and prevent preeclampsia continues to be poor.

Numerous biochemical markers have been studied for their ability to predict the onset of preeclampsia. Why do we seek a marker to predict preeclampsia when there is not a good treatment? It is hoped that if we could identify who was likely to develop preeclampsia then we could study interventions in that group which may ultimately lead to a way to prevent it. Unfortunately, no good markers have been identified as of yet.

In 2004 the World Health Organization did a systematic review of 7,191 potentially relevant scientific papers on this topic. Eighty-seven articles were ultimately included in the analysis and the WHO concluded that “As of 2004, there is no clinically useful screening test to predict the development of preeclampsia.”

However, also in 2004, Levine et al published a paper indicating that the circulating angiogenic factors called soluble fms-like tyrosine 1 (sFLT-1) and placental growth factor (PlGF) could be potential markers for the early prediction of preeclampsia. These proteins play a role in angiogenesis and are hypothesized to be required for normal embryonic vascularization. This caused a lot of excitement and led to many promising studies that examined the clinical utility of measuring Sflt-1 and PlGF to predict the onset of preeclampsia.

Unfortunately, these markers have not turned out to be all we had hoped they would be. In 2007 Widmer et al published the results from a systematic review of studies of sFlt-1 and PlGF. Ten of 184 available studies analyzing sFlt-1 and 14 of 319 studies analyzing PlGF were included in their review. The authors said that the evidence supports the possibility that sFlt-1 and PlGF are associated with the pathophysiology of preeclampsia or its phenotypes. In addition, third trimester changes in the blood concentrations of the markers were associated with preeclampsia, especially when the disease was severe. However, they concluded that “… the evidence is neither strong enough nor sufficient to recommend placental growth factor and sFlt-1 to screen women at risk to develop preeclampsia…" and "Prospective studies employing rigorous laboratory and study design criteria are needed to determine the clinical usefulness of these tests."

Apparently, there is a study called the "WHO Global Program to Conquer Preeclampsia" which was scheduled to start mid 2006; in this investigation approximately 10,000 women will be screened serially to evaluate these biomarkers for preeclampsia. So far, no data has been published on the results of this study…so stay tuned!

Determining a baby’s gender in the lab


Male Female It's rare that I come across a pregnant woman who doesn't already know the gender of the baby she is carrying.  Ultrasound assessment of a fetus is routinely performed during pregnancy to identify fetal anomalies and, very often, gender identification is offered.  After the 13th week of pregnancy, ultrasound is quite accurate in determining a baby's sex.  There are circumstances, however, where knowing a fetus is a boy or a girl is important.  For example, male fetuses born to women who are carriers of X-linked genetic disorders are at risk of inheriting the disorder.  In such instances, knowledge of a fetus' gender before 13 weeks can be advantageous.

Today, the Journal of the American Medica Association (JAMA) published a systematic review and meta-analysis (a systematic method of evaluating statistical data based on results of several independent studies of the same problem) that sought to determine the performance of DNA tests performed on maternal blood to determine fetal sex.  As a reminder, DNA from a fetus circulates in maternal blood.  This cell-free fetal DNA can be detected and amplified and is the basis of other emerging diagnostic tests.

This report evaluated the results from 57 different studies that tried to determine fetal gender using a test performed on a sample of mother's blood or urine.  The 57 reports contained results from 6,541 women; 54% who delivered boys and 46% who delivered girls.  Because this was a meta-analysis, several different testing techniques were used in the included studies but all of them looked for the presence or absence Y chromosome-specific molecular markers.  If those markers were detected, the fetus predicted to be a male.  If they were not detected, the fetus was predicted to be a female.

Collectively, the tests performed very well.  They correctly identified 95.4% of the male fetuses (test sensitivity) and 98.6% of the female fetuses (test specificity).  The tests did not work well if maternal urine was used as the test sample which is not surprising because cell-free fetal DNA is very difficult to detect in urine.  The report also noted that real-time quantitative PCR (RTQ-PCR) methods had better performance than conventional PCR techniques.

As might be expected, test performance improved along with the gestational age of the fetus.  At less than 7 weeks test sensitivity (see above for definition) was 74.5% but increased to 95% after 7 weeks at was 99.6% at more than 20 weeks.  That makes sense because there is more fetal DNA circulating in the mom's blood as the pregnancy progresses and the fetus increases in size.

Because of the potential clinical use of the test to identify male fetuses of mother's with X-linked genetic diseases, the authors caution that these types of sex-predicting tests should be optimized to achieve high sensitivity (detection of male fetuses) even if it compromises the specificity (detection of female fetuses).  They also cautioned that any tests developed and offered by laboratories should thorough validate that a negative (i.e. female fetus) test result reflects the presence of a female fetus and not simply the inability to detect any fetal DNA (which would also give a negative (female) result).

While there are some direct-to-consumer labs that offer this type of DNA-based testing to determine fetal sex, the test is marketed as one of convenience to satisfy a mother's curiosity regarding her baby's gender.  Currently, I am unaware of any clinical laboratories in the United States that offer such testing for diagnostic purposes although the tests are available for that reason in the European Union.  I predict that this type of test will become more widely available in clinical labs over the next few years.  There is much promise to be realized from tests based on circulating cell-free DNA!