The cerebroplacental Doppler ratio revisited


  • A. A. Baschat,

    Corresponding author
    1. Department of Obstetrics, Gynecology and Reproductive Sciences, University of Maryland, Baltimore, MD, USA
    • Department of Obstetrics, Gynecology and Reproductive Sciences, University of Maryland, 405 West Redwood Street, 4th Floor, Baltimore, MD 21201-1703, USA.
    Search for more papers by this author
  • U. Gembruch

    1. Department of Prenatal Medicine and Obstetrics, Friedrich-Wilhelm University, Bonn, Germany
    Search for more papers by this author



To evaluate the distribution of Doppler pulsatility index (PI) measurements of the umbilical and middle cerebral arteries in singleton fetuses of women with normal uterine artery blood flow and to construct reference ranges for the cerebroplacental PI Doppler ratio.


The PI was determined in the mid-portion of the umbilical artery and the mid- or distal segment of the middle cerebral artery in 306 normal singleton fetuses. The cerebroplacental Doppler ratio (CPR) was determined from paired measurements. After determination of the best fit, reference ranges were constructed for each parameter against gestational age (GA).


The PI for the umbilical artery had a linear relationship with GA (umbilical artery PI = − 0.0246 × GA + 1.7791, r2 = 0.4025, P < 0.001). The middle cerebral artery PI and the CPR both showed a quadratic relationship with GA (middle cerebral artery PI = − 0.0058 × GA2 + 0.3335 × GA − 2.7317, r2 = 0.2365, P < 0.01), (CPR = − 0.0059 × GA2 + 0.383 × GA − 4.0636, r2 = 0.2788, P < 0.001).


The CPR is not constant throughout gestation. Reference ranges constructed by a standardized Doppler technique may be of benefit in the monitoring of high-risk pregnancies. Copyright © 2003 ISUOG. Published by John Wiley & Sons, Ltd.


Doppler investigation of the fetal circulation can give important clues to fetal well-being in a number of fetal conditions. The unique arrangement of the fetal circulation allows afterload to affect ventricles individually. Accordingly, relative contributions of individual ventricles to the common cardiac output may change with individual alterations in afterload1. The blood flow changes that are most commonly applied to clinical management of fetal growth restriction are relative changes in cerebral and umbilical blood flow resistance. The brain-sparing effect, a decrease in middle cerebral artery blood flow resistance, was described over a decade ago2. In this setting improved oxygen delivery to the brain is thought to result from decreased left ventricular afterload. The cerebroplacental ratio (CPR) quantifies redistribution of cardiac output by dividing Doppler indices from representative cerebral and fetoplacental vessels3. The middle cerebral and carotid arteries, as well as the descending aorta and umbilical artery, have been used for this purpose4–6. Investigations in the lamb fetus suggest that the CPR most closely reflects acute changes in pO2 and therefore this ratio has been considered superior to the middle cerebral artery pulsatility index (PI)7 in predicting adverse fetal outcome4, 6.

There are a number of issues that merit consideration when using the CPR. Various vessels and sampling sites have been utilized. This has implications for the middle cerebral artery since the mid- and distal thirds have significantly higher PIs than the proximal third8. The CPR has been constructed using various Doppler indices (systolic/diastolic ratio, resistance index (RI), PI) and cut-offs (< 1, < 1.05, < 1.08) to predict adverse outcomes3–6, 9–12. Reference ranges are based on a wide range of patient numbers (20–1500 patients). Equally, both linear and quadratic regression models have been applied. These factors raise the concern of how the CPR should be constructed to best reflect the physiological gestational changes in both umbilical and middle cerebral artery Doppler indices. The aim of this investigation was to derive normative values for the CPR in uncomplicated pregnancies with normal maternal and neonatal outcomes.

Patients and Methods

The study was performed as a prospective cross-sectional analysis of Doppler measurements performed between 20 and 40 weeks' gestation in patients referred for ultrasound assessment of fetal growth. Patients were included in the final analysis if they met the following criteria: (1) normal fetal and neonatal anatomy, (2) accurate gestational age (GA) dating by either a first-trimester or early second-trimester (< 20 weeks) sonogram corresponding with the last menstrual period, (3) neonatal birth weight between the 10th and 90th percentiles for GA and gender and (4) normal uterine artery PI by local reference values.

Doppler examinations were performed by one examiner using 4- or 5-MHz sector transducers (Acuson Sequoia 512; Acuson, Mountain View, CA, USA) with spatial peak temporal average intensities below 50 mW/cm2 and the high-pass filter at 50–100 Hz. Measurements were obtained from the umbilical artery at the mid-section of the umbilical cord and the distal (straight) portion of the middle cerebral artery and the right and left uterine arteries by previously described methods2, 8. When at least five consecutive uniform flow velocity waveforms with a high signal-to-noise ratio were obtained during periods of fetal rest and apnea the image was frozen and the waveforms were quantified using the PI. For the uterine artery the presence of an early diastolic notch was noted. All patients included in the final analysis had both a normal uterine artery PI and waveform without early diastolic notching.

For patients with normal uterine artery Doppler indices and waveform pattern, paired Doppler results obtained from the umbilical and middle cerebral arteries were used in the final analysis. The CPR was derived as a simple ratio of the middle cerebral artery PI divided by the umbilical artery PI. The PI measurements of individual vessels and the CPR were examined for their distribution. Their relationship with GA was explored using regression analysis. After the best fit was determined, reference ranges and the 95% confidence interval (CI) were constructed for each parameter and displayed in graphic form. Statistical analysis was performed with SPSS 10.0 (SPSS Inc., Chicago, IL, USA). A value of P < 0.05 was considered statistically significant.


A total of 306 patients met the inclusion criteria and satisfactory waveforms and measurements were obtained in all cases (Table 1). The intraobserver variability for the pulsatility measurements as evaluated based on ten examinations was 5% throughout GA at the beginning and end of the study. Based on the variance of Doppler measurements from a 50-patient pilot sample it was calculated that at least 150 patients were necessary for the construction of reference ranges. The PI for the umbilical and middle cerebral arteries both showed significant relationships with GA. For the umbilical artery this relationship was best expressed by a linear regression (umbilical artery PI = −0.0246 × GA + 1.7791, r2 = 0.4025, P < 0.001). The relationship between the middle cerebral artery PI and GA was best expressed through a quadratic regression equation (middle cerebral artery PI = −0.0058 × GA2 + 0.3335 × GA −2.7317, r2 = 0.2365, P < 0.01). The CPR was not constant throughout and the relationship with GA was also best represented by a quadratic model (CPR = −0.0059 × GA2 + 0.383 × GA − 4.0636, r2 = 0.2788, P < 0.001). Reference ranges for the means and 95% CI were generated for each of these indices and are displayed in Figures 1–3.

Figure 1.

Graph showing the reference ranges (mean and 95% CI) of the umbilical artery pulsatility index.

Figure 2.

Graph showing the mean and 95% CI of the middle cerebral artery pulsatility index.

Figure 3.

Graph showing the gestational reference range (mean and 95% CI) of the cerebroplacental ratio based on paired measurements of the middle cerebral and umbilical artery pulsatility index.

Table 1. Pulsatility index in the umbilical and middle cerebral arteries and cerebroplacental Doppler ratio in 306 normal fetuses
Gestational weekNUmbilical arteryMiddle cerebral arteryCPR
  1. CPR, cerebroplacental Doppler ratio; SD, standard deviation.

37 60.950.242.060.682.250.66


This study evaluated the relationship between umbilical artery PI, middle cerebral artery PI and the CPR with GA. All of these variables are not constant but rather show a significant relationship with GA. For the middle cerebral artery and the CPR this relationship is best expressed by quadratic regression, while the umbilical artery PI shows a linear relationship with GA. Although these results are not novel, they call for a critical appraisal of reference ranges for the CPR that have been used.

Doppler ratios have been constructed using the descending aorta or the umbilical artery as an expression of right ventricular afterload and the carotid or middle cerebral arteries to reflect left ventricular afterload1, 2. The CPR appears to be more strongly influenced by abnormally elevated umbilical artery blood flow resistance, which may be due to the fact that placental arterial and venous blood flow changes may be interdependent affecting afterload and preload5. In addition, changes in blood flow resistance in the descending aorta and the umbilical circulation may be independent of each other4. In the cerebral circulation, examination of the middle cerebral artery is readily achieved at optimal insonation angles and therefore appears the preferred method in recent years. When examining the middle cerebral artery, the sampling site has significant effects on the Doppler measurements when the PI is used. Measurements from the middle and distal thirds have significantly higher indices than those from the proximal third8 but the proximal third appears to be less variable during changes in fetal behavioral state13. The clinical relevance of Doppler ratios constructed from different sites has not been evaluated to date.

Of the various Doppler indices, the RIs and PIs have been predominantly used to quantify both the umbilical artery and middle cerebral artery Doppler waveforms. The Doppler waveform is only represented on a scale from 0 to 1 when the RI is used although some authors use >1 if end-diastolic velocities are absent or reversed. In contrast, the PI provides ongoing waveform analysis over a wide range of waveform patterns and may therefore provide a more accurate representation of the downstream vascular resistance7. This is supported by several observations. When Doppler indices in various segments of the middle cerebral artery are compared, significant differences are only found when the PI is used8. In addition, reference ranges for the CPR which utilize the RI show a linear relationship with GA2, 3, 5, 11 unless large numbers are used12. In contrast, a quadratic relationship with GA is readily appreciated when the PI is used6. It is important to note the middle cerebral artery sample site when this index is used.

The CPR offers the advantage of detecting redistribution of blood flow due to two potential mechanisms. ‘Forced centralization’ that may be observed with elevated placental blood flow resistance1, 14 as well as decreasing cerebral blood flow resistance due to ‘brain sparing’2 can both affect the CPR. When compared to the individual vascular beds, the ratio shows greater variation and appears to offer earlier detection of fetal adaptation to placental insufficiency than either the umbilical or middle cerebral artery4. This concept has been explored to predict fetal and/or neonatal compromise using the CPR. Cut-off values that have been derived from small numbers of abnormal pregnancies3, 5, 9, 10 do not always show reproducible results15. In contrast, Bahado-Singh and coworkers documented improved prediction of adverse outcome in small-for-gestational age fetuses below 34 weeks' gestation when GA references for the CPR were used as cut-offs.

In prenatal Doppler surveillance, documentation of altered brain perfusion marks an important step in the fetal response to placental insufficiency. The CPR was described over a decade ago. However, non-uniform Doppler technique and variation in construction of normative data may limit the clinical utility of this potentially useful parameter. Further investigation is warranted to address these concerns.