Nuchal translucency and gestational age.

Professor Wald and his team (Wald et al., 2004) have remodelled the already modelled SURUSS data (Wald et al., 2003a) to claim, firstly, that their results are now similar to those of the multicentre study co-ordinated by the Fetal Medicine Foundation (Snijders et al., 1998), and, secondly, that the gestation of choice for nuchal translucency (NT) measurement is 10 weeks. The latest modelling exercise was apparently prompted by the study of Spencer et al. (2003a). However, the multiple of the median (MoM) approach used by Professor Wald for the analysis of the NT data is the same as the one shown by Spencer et al. (2003a) to be inappropriate. Furthermore, in SURUSS, there were 101 fetuses with trisomy 21 but NT was measured in only 85 (Wald et al., 2003a). At 10 weeks, there were six trisomy 21 fetuses and the NT was below the 95th centile in all the cases. In the analysis, which now claims a 73% detection rate, for a 5% false-positive rate, the data from only three of the six cases were used. We summarise the results of the study by Spencer et al. (2003a). In screening for trisomy 21 by NT, patient-specific risks are derived by multiplying the a priori maternal age and gestation-related risk by a likelihood ratio, which depends on the difference in fetal NT measurement from the expected normal median for the same fetal crown-rump length (delta value). In screening using maternal serum–biochemical markers, a different approach has been used to take into account the gestation-related change in marker levels. This method involves converting the measured concentration into a multiple of the median of unaffected pregnancies at the same gestation. The study of Spencer et al. (2003a), involving analysis of data from 128 030 unaffected and 428 trisomy 21 pregnancies, demonstrated that the delta NT approach provides accurate patient-specific risks. In contrast, the traditional MoM approach was found to be inappropriate for this purpose because none of the three basic assumptions that underpin this method are valid. Firstly, in the unaffected population, the distributions of NT MoM and log10 (NT MoM) were not Gaussian; secondly, the SDs did not remain constant with gestation; and thirdly, the median NT MoM in the trisomy 21 pregnancies was not a constant proportion of the median for unaffected pregnancies. The NT MoM approach resulted in women being given an overestimate of risk for trisomy at 11 weeks and a considerable underestimate of risk at 13 weeks (Spencer et al., 2003a). The non-Gaussian nature of Log10 (NT MoM) still persists for gestation specified by week or by day, so the calculation of detection rates and falsepositive rates on the basis of this assumption needs to be carefully verified. In screening for trisomy 21 by fetal NT, the likelihood ratio associated with delta NT is effectively constant for gestations between 10 and 13+6 weeks (Spencer et al., 2003a). Professor Wald’s letter also presents the effect of the remodelled data of SURUSS on the performance of the integrated test. However, there are concerns about both the firstand the second-trimester biochemical components of this test. Firstly, Professor Wald’s claim that 10 weeks is the best gestation for starting the process of screening is essentially based on the good performance of PAPP-A. However, in SURUSS, the data on PAPP-A from 10 weeks were combined with those at 9 weeks, which would exaggerate the importance of this metabolite. In particular, the contribution of samples from 9 weeks was 16% for the unaffected pregnancies and 40% for the trisomy 21 pregnancies (Wald et al., 2003a). Secondly, in SURUSS, the predicted detection rates, for a 5% false-positive rate, were 71% for the double test, 77% for the triple test and 83% for the quadruple test (Wald et al., 2003a). These detection rates are substantially higher than the respective rates of 61, 66 and 75% reported by the same authors in their prospective screening studies (Wald et al., 2003b). Thirdly, Spencer et al. (2002; 2003b) have demonstrated temporal changes in maternal serum–biochemical markers of trisomy 21 across both the first and second trimesters of pregnancy. Consequently, calculation of accurate patient-specific risks requires a variable median–separation model (Spencer et al. 2003b), rather than the conventional constant median–separation model proposed by Wald et al. (1988). Fourthly, there are major concerns on the robustness of the inhibin A assay. In 1992, we selected 10 weeks as the earliest gestation for measurement of NT (Nicolaides et al., 1992) because screening necessitates the availability of a diagnostic test and chorionic villous sampling before 10 weeks may be associated with transverse limb–reduction defects (Firth et al., 1991). We subsequently changed the minimum gestation to 11 weeks because it was realised that at this gestation, but not at 10 weeks, many major fetal abnormalities, such as anencephaly, cardiac defects, obstructive uropathy, exomphalos and some cases of spina bifida, can also be diagnosed at the NT scan. Advances in fetal medicine should not be reversed by remodelling of modelled detection rates for trisomy 21.