Trisomy 21, also known as Down Syndrome (DS), was the most common chromosomal abnormality, which were accompanied by different degrees of mid-facial hypoplasia and skin edema[12, 13]. DS fetuses had typical facial features compared to euploid fetuses, including hypoplastic or absent nasal bones, thickened prenasal skin, shortening and dorsal displacement of the maxilla, et al. Most fetuses with trisomy 18, the second most common chromosomal abnormality, had micrognathia and CLP. The diagnosis of micrognathia was mainly subjective during second and third trimester. In the meanwhile, CLP was the most common facial malformation. Although it was not fatal, it had great adverse impact on children and families involved. Moreover, 54% of CLP may associated with other anomalies or genetic syndromes , affecting about 1 ~ 2/1000 live births. In this study, multiple facial markers that reflected the relative position of the forehead, maxilla, and mandible were analyzed to establish their normal reference ranges, and these markers could further provide objective and quantitative criteria for the early detection of fetal facial anomalies and underlying genetic abnormalities.
The embryonic development of facial bones has its main characteristics. The maxilla and mandible begin to ossify from 8 weeks onward. The maxilla is anatomically fused with the skull and grows forward with the development of brain tissue, while the mandible is connected to the skull through the temporomandibular joint. Therefore, the mandibular forward growth rate during first trimester is slower than that of the maxilla. From 20 weeks onward, the maxilla ossification has almost completed, and the developing fetal swallowing function accelerates the growth of mandible. After that, the position of facial bones is relatively constant, then fetal facial profile is basically formed .
In 2002, Rotten et al.  first introduced the IFA to detect micrognathia. They found that the mean IFA of normal fetuses was 65.5°±8.13° at 18 ~ 28 weeks’ gestation, and it was constant during pregnancy. Using 49.2° (average-2SD) as a cut-off point, the IFA had a sensitivity of 1.0, a specificity of 0.989 to predict micrognathia. IFA could reflect the anterior and posterior position between the mandible and frontal bone to evaluate micrognathia, which was often associated with some genetic anomalies, such as Pierre-Robin syndrome, Stickler syndrome, trisomy 18 and trisomy 13[1, 21, 22]. During first trimester, we found that the mean IFA of normal fetuses was 80.2 (SD 7.25) °, and it decreased with CRL. This value was larger than Rotten et al. . The reason might be that the mandibular forward growth rate during first trimester is slower than that of the forehead. While during second trimester, the position of facial bones is relatively constant. In our study, the reference range of IFA was 65.99°~94.41°. When IFA was less than 65.7° (average-2SD), the possibility of micrognathia should be considered, which was helpful for the early detection of certain genetic syndromes. However, the clinical significance of this cutoff value needed to be confirmed by large abnormal sample from multi-centers.
The MNM angle could reflect the relative position of the maxilla and mandible, further to evaluate fetal facial profile. De Jong-Pleij et al.  reported that the mean MNM angle was 13.5° and did not change in second and third trimester, which was a sensitive indicator for evaluating micrognathia and CLP. Vos et al. [23, 24] reported that the MNM angle had a definite implication for trisomy 21 and trisomy 18. In our study, the mean MNM angle was 4.17 (SD 1.19) °, which increased with CRL. It was close to Ko et al.  study (4.7°±3.3°) but smaller than the Lu et al.  study (12.4°±2.2°). The reason might be racial differences. During first trimester, the maxilla and frontal bone are directly connected with the skull, the mandible grows forward more slowly than the maxilla and frontal bone, which may cause the MNM angle to increase with CRL. While during the second trimester, the ossification of the maxilla completes, and the development of fetal swallowing function accelerates the growth of the mandible and the formation of fetal facial profile, so the MNM angle does not change with gestational age. Further studies are necessary to investigate the relationship between the MNM angle and fetal facial abnormalities or chromosomal abnormalities.
In order to avoid the influence of the curvature of the vomer, Lu et al.  used the surface of anterior half of the maxilla as a reference line. FMA could directly reflect the relative position of the maxilla and mandible and be independent of other facial structures. Their research showed that FMA was related to gestational week, which increased with gestation slightly (1°~ 2°/week) from 16 weeks till 28 ~ 31 weeks and decreased minimally thereafter. It might be consistent with the allometric growth relationship between different parts of fetal face. We found that FMA increased with CRL, with the reference range of 64.95°~85.77°. Lu et al.  reported the cut-off value of FMA in detecting micrognathia was 66°, with the detection rate of 100% and false positive rate of 2.5%, which was similar to our study (64.95°). A large prospective cohort was needed to determine the diagnostic accuracy of FMA for micrognathia during first trimester.
De Jong-Pleij et al.  showed that the mean PL distance at 27 ~ 36 weeks’ gestation was 2.8 (range 2.1 ~ 3.6) mm, and 4 mm could be used as the upper limit of the normal for judging frontal bossing. The PL distance was the first objective quantitative indicator to assess frontal bossing, which was affected by the position of the mandible, nasion and frontal bone. In our study, the mean PL distance was 2.78 ± 0.54 mm, and it decreased with CRL, which was consistent with Bakker et al. . It might be caused by a forward movement of the maxilla and a decrease in convexity of the forehead during first trimester. Bakker et al.  also pointed out that the PL distance was not the best ultrasound marker for aneuploidies.
There were some limitations in our study. Firstly, static image randomly selected from first trimester ultrasound screening was to measure the NT thickness, not specifically to observe facial abnormalities. Secondly, all parameters were measured on 2D images, without the use of 3D reconstructed techniques. Some research [7, 14] showed that 3D technique could better obtain the true mid-sagittal section, but it took a long time. On the other hand, 2D ultrasound was the basis of 3D ultrasound. 2D measurements were reported to be the same reliable and accurate as 3D measurements in the measurement of facial marker[27, 28].