Fetal nasal bone length: reference range and clinical application in ultrasound screening for trisomy 21
MATERIALS AND METHODS
This prospective screening study was conducted in theFetal Medicine Unit of the University of Sao Paulo MedicalSchool Hospital from October 1993 to December 1999.The length of the fetal nasal bone was measured in patientsaged at least 35 years as part of a multiple-parameterultrasound-screening program for trisomy 21.
The study population included 1923 women agedbetween 35 and 45 years and with pregnancies ofgestational ages varying between 16 and 24 weeks andof unknown fetal karyotype. The main indications for theanomaly scans were: routine policy, option for ultrasoundscreening after parental decision to avoid having aninvasive test, or immediately before amniocentesis.
The following criteria were used for exclusion: knownkaryotype, previous history of chromosomal abnormalities, referral for previously suspected ultrasound anomaly,or fetal death diagnosed at ultrasound examination.
The fetal nasal bone length was measured in a strictsagittal view of the fetal head, identifying the nasal bone,lips, maxilla, and mandible with an angle between theinsonation beam and the nose axis of close to 45◦. Themaximum length of the nasal bone was measured inmillimeters to one decimal place. After identification ofthe appropriate plane, three independent measurementswere obtained and averaged. The correct plane for themeasurement of fetal nasal bone length is illustrated inFigure 1.
Gestational age was calculated in decimal weeks bymeans of the last menstrual period in patients withknown and regular 26–30-day periods. Whenever thesecriteria were absent and significantly discordant fetalbiometry was found, gestational age was estimated byfirst-trimester biometry (crown–rump length) or twosubsequent second-trimester scans (head circumference)
In cases where more than one anomaly scan wasperformed per patient, only the first scan results wereused for this analysis. The scans were performed by oneof four qualified operators, under direct supervision ofthe first author, using one of four different machines:Toshiba SSA 140 A (Toshiba, Otawara, Japan), ToshibaSSA 320 A – ECCO Cee (Toshiba, Otawara, Japan),Ultramark 9 (Advanced Technology Laboratories, BothellWA, USA) and Siemens Quantum 2000 (Siemens,Munich, Germany).
Follow-up was obtained in each case either by letter,by telephone contact or by karyotype results wheneveran amniocentesis was performed. Informed consent wasobtained and the research project was approved andmonitored by the hospital ethics committee.
Statistical analysis
Following data collection, statistical analysis was performed using SPSS for Windows v.10.0 (Chicago, IL,USA) software package. A scatter plot for fetal nasal bonelength measurements as a function of gestational agewas obtained and mean, 95% confidence intervals and5thpercentile were estimated by least-squares regression.The methods used for constructing the normal range aredetailed elsewhere8,9.
Diagnostic power to detect trisomy 21 of nasal bonelength measurements shorter than the 5thpercentilefor the gestational age was calculated by means of a2×2 table. Comparisons between normal and Downsyndrome fetuses were conducted by Fisher’s exact test,and sensitivity, false-positive rate and likelihood ratiowere computed.
RESULTS
Maternal age varied from 35 to 45 years (mean, 38.5)and showed no correlation with fetal nasal bone lengthmeasurement.Follow-up was possible in 1631 cases (84.8%). Trisomy21 was found in 22 cases resulting in an overall incidenceof Down syndrome of 1.35% in this population. Innine cases other chromosomal abnormalities were found(three trisomies 18, two trisomies 13, two triploidies, andtwo 45,X) and the affected fetuses were excluded fromthe study.
The mean gestational age was 20.4 weeks. Thegestational age was calculated by the last menstrualperiod in all the trisomy 21 cases and in 1042of the chromosomally normal fetuses. In a further348 cases the gestational age was calculated by thefirst-trimester crown–rump length measurement and in219 cases gestational age was calculated by secondtrimester biometry.
Nasal bone length measurement showed a significantincrease with gestational age (P<0.05). Mean nasal bonelength measurement was 6.9 mm and the overall standarddeviation was 1.29 mm.
Nasal bone length measurement showed a significantincrease with gestational age (P<0.05). Mean nasal bonelength measurement was 6.9 mm and the overall standarddeviation was 1.29 mm.
The increase of the nasal bone length accordingto the gestational age was given by the equation:NBL=0.27×GA+1.41, where NBL is the nasal bonelength (mm) and GA is gestational age (in decimalweeks). The linear model was preferred since higherorder polynomial coefficients of the equation were notsignificantly different from zero or only a marginally betterfit to the data was obtained8,9. Study of the residualsdiscarded heteroscedasticity, therefore allowing the useof a constant standard deviation around the regressionreference range. The standard deviation of the residualswas 1.13 mm.
The resulting reference ranges are graphically illustratedin Figure 2. Fifth percentile cut-offs for each gestationalage are presented in Table 1.
The nasal bone length was below the 5thpercentilein a significantly higher (P<0.05) proportion of fetuseswith Down syndrome than in normal fetuses. In 13 of22 (59.1%) cases of trisomy 21 and in 82 of 1600(5.1%) normal cases the measurement was below the
5thpercentile. Figure 3 illustrates an abnormally shortnose measurement in a fetus with Down syndrome.
Screening for trisomy 21 using the 5thpercentile as acut-off value resulted in a sensitivity of 59.1% for a 5.1%false-positive rate and the likelihood ratio was 11.6.
DISCUSSION
Women aged 35 years or more in current prenatal careare considered to be at risk for fetal Down syndrome andare therefore routinely offered invasive tests in order torule out chromosomal abnormalities.
Some of these patients prefer to have a screeningtest for Down syndrome because of the risk of fetalloss associated with invasive diagnostic procedures.Cases eligible for second-trimester sonographic screening include: patients who have not undergone a first-trimesterscan, those opting for sequential screening followingnuchal translucency measurement in the first trimesteror those undergoing second-trimester maternal serumbiochemistry screening.
Second-trimester sonographic screening is based onmultiple morphological and biometric parameters, including major structural anomalies, increased nuchal skin fold,pyelectasis, short humerus and femur length, and hyperechogenic bowel5. The incidence of chromosomal defectshas been shown to increase dramatically with the numberof sonographically detected abnormalities10.
Facial features associated with trisomy 21 includeslanted palpebral fissures, an epicanthal fold, a flatprofile, a protruding tongue, and a wide and saddle-likenose7. Despite the fact that prenatal ultrasound diagnosisof these abnormalities is possible, the subjectiveness ofthese markers has prevented their clinical application insonographic screening for Down syndrome.
Goldsteinet al.11and Pinetteet al.12studied the growthof the fetal nose width and nostril distance in normalpregnancies. The latter group found that these distanceswere above one standard deviation of the mean in 80%of trisomy 21 fetuses. Nonetheless, this cut-off resulted inan unacceptably high false-positive rate (33%).
Recently, Ciceroet al.13found that at the 11–14-weekscan the nasal bone was absent in 43 of 59 (73%) casesof trisomy 21, while it was absent in only three of 603(0.5%) chromosomally normal fetuses. The authors founda likelihood ratio of 146.0 for an absent nasal bone. Theseresults still need to be validated in a prospective study inorder to be applied in the clinical setting, as they could bebiased by a high prevalence of trisomy 21 (8.4%). If thesedata are confirmed, the impaired ossification of the fetalnose will prove to be one of the most specific markers for trisomy 21.
Guiset al.14were the first to present a referencerange for the growth of the nasal bone from 14 to34 weeks. They found that the measurement of thefetal nasal bone in a strict sagittal facial plane isfeasible and reproducible, increasing consistently withgestational age. Nevertheless, the small number of cases(376 measurements) resulted in relatively wide confidenceinterval reference ranges. Furthermore, no comparisonwas made between nose length in Down syndrome fetusesand that in chromosomally normal fetuses.
Sonek and Nicolaides15have recently reported threecases of trisomy 21 in the mid-trimester, in which thenasal bone was absent or shorter than expected for thegestational age. They concluded that the measurementof the fetal nose might improve our ability to detectDown syndrome.
The present paper presents confidence intervals for reference ranges for the measurement of the fetal nasal bonelength in the second trimester in a large population set(1600 cases) and thus provides useful information for theclinical application of the measurement in screening fortrisomy 21.
Using this reference range in a prospectively screenedpopulation, it was found that a measurement of the fetalnasal bone that falls below the lower limit (5thpercentile)shows a high likelihood ratio (11.6) for trisomy 21,since abnormal fetuses had a significantly smaller nasalbone than did those with normal karyotype (P<0.05).Therefore, the subjective impression of a flat profileobserved in fetuses with Down syndrome was objectivelyconfirmed by this method.
As an isolated screening test, the nose length has showna satisfactory performance, comparing favorably withother isolated parameters described in the literature. Infact, the sensitivity of this method (59.1%) for a 5% falsepositive rate was similar to that of other well-establishedscreening strategies, such as maternal serum biochemistry.
Theoretically, the detection rate could be even higher ina low-risk population since ‘risk screening’ would be usedto combine maternal age with the nasal bone likelihoodratio. The finding of a short nose was shown to increasethe background risk by 11.6 times. This means that for aninitial risk of trisomy 21 of 1 : 500 for a given maternaland gestational age, a short nose would mean that the riskis raised to as high as 1 : 43, the same risk as for a mucholder woman.
Ideally, the nose length could be combined with thematernal age, nuchal skinfold thickness and other softmarkers and even with biochemistry screening findings.However, independency of all these variables must bedemonstrated before this approach is proposed.
One of the potential limitations of the method is thepresumed high variability of the fetal nasal bone lengthwithin different races. In this study, the racial variabilityissue was not addressed due to the high miscegenationfound in Brazil. Racial factors may partially explainsome of the discrepancies between our reference rangesand those reported by Guiset al.14, which includedonly Caucasians.
Another source of concern about the measurement ofthe fetal nasal bone is repeatability. Data from Guiset al.14showed little interobserver and intraobserver variability. Notwithstanding, a strict facial profile is necessaryfor a correct nose measurement14, which requires substantial sonographer training. Furthermore, the angle betweenthe ultrasound beam and the axis of the nasal boneshould be close to 45◦. Incorrect angles are a potential source of error, with sharp angles resulting in atendency to underestimate the measurement and anglesapproaching 90◦making edges of the nasal bone moredifficult to be delineated15. Further studies demonstratingfetal nose measurement reproducibility and repeatabilitywill be decisive for its incorporation into routine clinical screening.
A final challenge to the method is the feasibility ofobtaining a facial profile. The mid-sagittal view of the facedemonstrating the fetal profile can be obtained within 2–3minutes in approximately 75% of cases16,17. This view ismost easily achieved with the fetal head in the transverseor occiput posterior position. On the other hand, the‘back-up’ position of the fetal head makes measurement
of the nasal bone more difficult and time-consuming.Diminished amniotic fluid and maternal obesity may alsoimpair fetal face visualization.
While it is expected that a transvaginal approach andthree-dimensional techniques could save time or enhancefetal face evaluation in selected cases18, we have foundthat visualization and measurement of the nasal boneby experienced operators was virtually always feasibleusing conventional two-dimensional sonograms, which isin agreement with the results of Guiset al.14.
Further studies are needed to study racial variability andto validate the method for use in younger women and aspart of sequential strategies associated with first-trimesterscreening. Validation of this model with a larger numberof Down syndrome cases would also allow estimationof the multiple of the median adjusted likelihood ratiosfor each gestational age and nasal bone measurement,which would certainly improve sensitivity and reducefalse-positive rates.