Combined first trimester nuchal translucency and second trimester biochemical screening tests among- normal pregnancies

A. Herman (E-mail: aherman@asaf.health.gov.il), Z. Weinraub, E. Dreazen, S. Arieli, S. Rozansky, I. Bukovsky and R. Maymon

INTRODUCTION

The introduction of second trimester biochemical screening was a milestone in the process of antenatal detection of Down syndrome (DS) in a low-risk population (Merkatz et al., 1984; Cuckle and Wald, 1984).

The model for assessing a combined risk for DS published over a decade ago by Cuckle et al. (1987) is still valid today and has been broadened to incorporate additional markers (Wald et al., 1988).

Based on the risk calculated from the test, patients and health providers are afforded more information to assist in decision making on invasive fetal karyotyping.

A first trimester ultrasound screening test was recently introduced into prenatal care (Snijders et al., 1996, 1998; Taipale et al., 1997; Hafner et al., 1998).

Based on the measurement of nuchal translucency (NT) thickness, this method was evaluated by a study group of the Royal College of Obstetricians and Gynaecologists (1997) and was judged as being 'an acceptable procedure' for DS screening.

Although some authors utilized a cut-off value of 2.5 mm (Hafner et al., 1998) or 3.0 mm (Taipale et al., 1997), the recent trend is to generate a calculated risk based on mathematical models which combine maternal age related background risk, actual NT thickness and gestational age (Snijders et al., 1996; Biagotti et al., 1997).

Both first trimester ultrasound and second trimester biochemical screening tests are associated with similar screen-positive rates of about 5% (Wald et al., 1988; Snijders et al., 1996, 1998; Hafner et al., 1998) and a sensitivity of 50-60%, (Wald et al., 1988, 1992; Taipale et al., 1997; Haddow, 1998; Haddow et al., 1992).

Some studies reported that the introduction of first trimester screening led to decreased efficacy of second trimester screening (Kadir and Economides, 1997; Thilaganathan et al., 1997).

Wald el al. (1999) recently presented the potential benefit of integrating first and second trimester screening tests, reaching a detection rate of 85% and a screen-positive rate of 1%.

Those authors cited Larn et al. (1998) who reported on lack of correlation between multiple of medians of NT with alpha-fetoprotein or human chorionic gonadotrophin.

Yet, the editorial that accompanied that paper (Copel and Bahado-Singh, 1999) noted that since that study used published data of different groups - without knowledge of the overlap - their results required validation.

The aim of the present study was to esup or equal toamine the overlap and degree of correlation between first trimester ultrasound and second trimester biochemical screening tests (taken as a whole), among normal pregnancies, and to evaluate the effect of various combination methods on the screen-positive rates.

PATIENTS AND METHODS

A total of 750 women underwent first trimester ultrasound screening at the Assaf Harofeh Medical Center during 1997.

Since the present work was designed to study singleton pregnancies at low risk for aneuploidy, 112 cases were esup or equal tocluded for the following reasons: 52 multiple pregnancies, 50 patients > or = 35 years of age, six missed abortions and four cases with major anomalies.

Of the remaining 638 pregnancies, there were 16 cases with increased NT thickness resulting in a DS risk > or = 1: 380 (2.5%).

Second trimester biochemistry tests were not performed in sisup or equal to of these 16 cases because five undewent early karyotyping (three had a normal karyotype, one had trisomy 18 and one had Turner's syndrome) and one decided to terminate the pregnancy (the abortus material showed normal karyotype).

Information on second trimester screening, invasive procedures and pregnancy outcome was obtained either from hospital records or by a telephone interview.

Ninety-seven patients cou]d not be located (15%).

Of the 77 patients who did not undergo second trimester screening, 15 were referred directly to amniocentesis, eight reported on spontaneous abortion (karyotype unknown) and four refused on religious grounds.

The final study cohort consisted of 508 singleton pregnancies defined a priori as being at low risk for aneuploidy and for which the results of both screening tests were available.

All patients were followed until delivery and there were no indications for karyotyping throughout their pregnancies.

Screening tests

First trimester ultrasound screening was carried out between 10-14 weeks' gestation in our facility using a transabdominal curvilinear transducer (Advanced Technological Laboratories, HDI 3000, Seattle, WA, USA).

The scans were performed by qualified esup or equal toaminers (AH, ED and RM) and NT thickness was measured according to the method previously described by Nicolaides et al. (1992).

The measurements and scans underwent esup or equal toternal and internal audit (Snijders et al., 1996; Herman et al., 1998).

The programme was approved by our local hospital authority and all patients signed an informed consent form.

Results of DS screening tests are presented in Israel in terms of risk for DS livebirth.

At the end of each esup or equal toamination, the patient was counselled, and the risk for DS livebirth was presented to the couple (based of software supplied by the Fetal Medicine Foundation) (Snijders et al., 1996).

A risk of > or = 1: 380 was considered screen-positive and fetal karyotyping was recommended, regardless of later results of biochemical screening.

Second trimester biochemical screening was performed between 16-19 weeks' gestation in various Ministry of Health approved laboratories in the community.

The esup or equal toamination included the three components of the 'triple test' (TT) i.e. alpha fetoprotein, human chorionic gonadotrophin and unconjugated oestriol.

The same cut-off level for DS livebirth (risk > or = 1: 380) was used for defining cases as screen-positive, and all the women with this risk were advised to undergo prenatal karyotyping regardless of the results of first trimester ultrasound screening.

Correlation and overlapping between tests

In order to evaluate the net correlation between the results of NT and TT (taken as a whole), maternal background risks were eliminated by using their likelihood ratios (LR = calculated risk/background risk).

The Spearman rank correlation test was used before and after logarithmic (log) transformation of the LRs (Orlandi et al., 1997).

McNemar's paired Chi-square test was employed to assess different sreen-positive rates obtained from NT and from TT at a cut-off level of > or = 1: 380 of DS livebirth risk.

To demonstrate the actual interrelationship between the tests, a histogram demonstrating the mutual distribution of the logarithmic transformation of the LRs was plotted.

Combined methods

For assessment of the screen-positive rates of both tests in combination, we exmined two combined methods :

• the 'summation risk' which includes all screen-positive cases (DS risk sup or equal to 1: 380) by either test (avoiding repetitions),
• and the 'calculated risk' which involves the generation of a new risk derived from a combination of LRs (LRcombined=LRnt sup or equal to LRtt) (Orlandi et al., 1997; Cuckle and Sehmi, 1999).

Both LRs amplified each other when they were in the same direction (> 1 or < 1).

When one was close to 1, the other became dominant and when both were in opposite directions, the integrated risk attained an intermediate value.

Amniocentesis rates were examined according to various age groups ( < 24, 25-29 and 30-34 years).

RESULTS

The likelihood ratios (derived from NT and TT screening tests) and their log transformation did not correlate (Spearman rank corre]ations of 0.009, p=0.8 and 0.04, p=0.3, respectively).

At the cut-off level of > or = 1: 380 there were 468 cases defined as screen negative by both tests, nine as screen-positive by NT only, 30 as screen-positive by TT only and one as screen-positive by both tests (p<0.001, McNemar's paired Chi-square test).

This means that, in normal pregnancies, while the result of one of the tests showed an increased risk for DS livebirth, the probability that the result of the other test would do the same was negligible.

At this cut-off level, the screen-positive rate derived from NT (2%) was lower than that derived from TT (5.7%).

Categorization of the patients into age groups showed that screen-positive rates actually increased parallel to increased maternal age (NT: 0%, 1.3% and 3.3% and TT: 1.7%, 2.9% and 10%, for age groups < or = to 24, 25-29 and 30-34 years, respectively).

Table 1 shows a substantial difference of screen positive rates, using different combining methods.

Those were highest for the summation risk, reaching 7.5%, and those obtained by calculated risk were as low as 2.0%.

The same pattern of increased screen positive rates parallel to the increase in maternal age was demonstrated by both methods.

Figure 1 presents the mutual distribution of the patients according to the log transfomation of LRs.

As expected, most of the cases were concentrated at the lower values, pointing towards mutually decreased risk.

There were only two pregnancies in which both the LRnt and LRtt demonstrated an increased DS risk (LR > I, represented by log LRs > 0).

In 11 out of the 14 cases exibiting increased risk by NT (log LRNT > 0), the results of the corresponding TT showed a decreased risk (LR < 1 represented by log LRTT < 0).

Similarly, in 43 out of 46 cases exhibiting an increased risk by the TT (log LRtt > 0), its corresponding NT results showed a decreased value (log LRnt < 0).

Thus, in 55 out of the 58 (95%) cases exhibiting an increased risk by either NT or TT, the corresponding test demonstrated a decreased value.

In our study cohort with an a priori low risk, 20.7% of the women underwent amniocentesis.

While 7.5% underwent 'indicated' amniocentesis (screen-positive by either test), 13.2% opted for a non-indicated karyotyping (screen-negative by both tests).

This was more frequent in the 30-34 year age group, in which 32.4% underwent amniocentesis of which 20% were not 'indicated'.

DISCUSSION

The approach which combines the results of different DS screening tests relies on the assumption that the tests are independent determinants of DS risk (Cuckle and Selmli, 1999).

Specifically for NT and the biochemical tests, this assumption is based on studies which could not find a correlation between NT and each of the first or second trimester serum markers (Orlandi et al., 1997, Lam et al., 1998).

The present study attempts to be the first to show that, among normal pregnancies, the results of NT and TT (taken as a whole) indeed do not correlate and thus validate the concept of combining their results.

Of the initial 638 singleton pregnancies, there were 16 (2.5%) cases with an increased risk, which was derived from the NT examination.

However, for the six women who elected to have early karyotyping or termination of pregnancy, we could not determinine the combined risk.

We propose that first trimester screen-positive cases, electing to have second trimester karyotyping, should be followed closely and encouraged to have the biochemical test, as did 10 of our study women.

We used the cut-off risk of 1: 380 for DS livebirth as a reference, based on the value cited by the Israeli Ministry of Health as an indication for karyotyping in patients who are younger than 35 years at the beginning of pregnancy (Legum el al., 1994).

Because the odds for DS vary considerably between first trimester, second trimester and term (Snijders et al., 1994), the risks obtained at various time periods, or by different tests, should be adjusted to a common denominator to be taken as the risk for livebirth of DS at term.

The advantage of a method which combines the results of different tests and yields a single value is obvious.

Such a method was recently developed for combining first trimester ultrasound and biochemistry (Wald and Hackshaw, 1997).

In the current study, we tried to evaluate the effect of combining two methods in terms of screen-positive rates: a simple sumunation of the different screen-positive pregnancies resulted in the highest result, reaching 7.5%.

The calculated combined risk utilizes the same principles applied when combining results of different biochemical analytes and is, indeed, associated with a much lower screen-positive rate of 2%.

The advantage of such a low rate should be evaluated in terms of a comprehensive detection rate which is currently not available.

We demonstrated negligible overlapping between NT and TT screening tests in screen-positive cases.

At a cut-off level of > or = 1:380, there was only one case identified as screen-positive by both tests (2.5% overlapping).

Although our series is relatively small in number, the lack of correlation between the LRs was highly significant.

Moreover, of 508 cases, there were only two which exhibited a LR > 1 and only six in which both LRs were > 1.

Thus, the addition of another few more cases would have no effect on the results.

An indirect support of our observation can be found in a recent study by Kadir and Economides (1997) who reported a second trimester screen-positive rate of 9.6% one year prior to NT screening implementation and 10% thereafter.

The unchanged false-positive rate of the biochemical screening that they reported—despite the exclusion of NT screen positive cases—suggests minimal overlapping.

Since that type of data were not obtained for DS affected pregnancies, the concept that NT and screening tests complement each other and, taken together, may identify most of DS cases warrants further confirmation.

Once it is validated, wonmen could be reassured of having an actual Iow DS risk in cases defined as screen-negative by both tests, hopefully leading to a decrease in the number of non-indicated invasive procedures.

In summary, our findings validate Wald et al.'s (1999) theoretical calculations that integrating NT and TT results in a low screen-positive rate.

We have shown that, among normal pregnancies, the results of NT and TT are not correlated and that the overlapping of screen-positive cases is minimal.

These screening tests complement each other and, in combination, may identify more DS cases as well as reassure women who screen negatively by both tests.

More studies are needed to delineate the optimal method of combining the tests with the goal of reducing the number of invasive procedures while simultaneously detecting DS cases with high efficiency.

Mise en page pour ECHOLIST le 27/12/2000 par le Dr JM BRIDERON (jmbr@wanadoo.fr)