stroke prevention in sickle cell disease (stop) study guidelines for transcranial doppler testing

Journal of Neuroimaging Vol 11 No 4 October 2001Nichols et al: STOP Study Guidelines for TCD Testing

Views and Reviews

Stroke Prevention in Sickle CellDisease (STOP) Study Guidelinesfor Transcranial Doppler Testing

Fenwick T. Nichols, MD

Anne M. Jones, RN, BSN, RVT, RDMS

Robert J. Adams, MD

A B S T R A C T

The Stroke Prevention in Sickle Cell Disease (STOP) trial usedtranscranial Doppler (TCD) to screen children with sickle cell dis-ease with no history of stroke. Children (who consented) whohad time-averaged mean of the maximum (TAMM) velocities inthe middle cerebral artery and/or distal internal carotid arterywere randomized to transfusion or standard. Over a slightlymore than 20-month average follow-up, there were 11 strokesin the standard care arm and 1 stroke in the transfusion arm.This study has caused a great deal of interest in using TCD toscreen children with sickle cell disease. For the STOP TCD datato be applied appropriately, it is necessary for users of TCD tounderstand how the STOP TCD examinations were performed,how the TCD velocities were measured, and which velocitieswere used. This article will review the STOP TCD scanning pro-tocol and the reading protocol and review the TAMM velocityand how it differs from other velocity measurements.

Key words: Sickle cell anemia, ultrasonography, Doppler,transcranial, childhood, adolescence, blood flow velocity.

Nichols FT, Jones AM, Adams RJ.Stroke prevention in sickle cell disease (STOP) study

guidelines for transcranial Doppler testing.J Neuroimaging 2001;11:354–362.

The Stroke Prevention in Sickle Cell Disease (STOP)study,1-3 which screened children with sickle cell diseasewho had never suffered a stroke, demonstrated thattranscranial Doppler (TCD) could be used to identify chil-dren who were at high risk for development of stroke.This study also demonstrated that if these high-risk chil-dren were treated with transfusion to maintain the hemo-globin S at less than 30%, their risk of developing a strokewas decreased by 90% when compared to the high-riskgroup that received standard care. As a result of thesefindings, there has been significant interest in the ultra-

sound community in using Doppler ultrasound to screenchildren with sickle cell disease. This article will brieflyreview the differences between routine clinical TCD andTCD as it was used in STOP so that Doppler users will beable to understand some key issues related to STOP. Thisarticle will review the STOP examination and readingprotocols and address specifically the issues of the TCDvelocities used to determine stroke risk.

Background

Children with sickle cell disease (hemoglobin SS) have asignificant risk of developing ischemic stroke, with 11% ofHbSS patients suffering a stroke before the age of 20.4

These strokes are primarily the result of stenosis or occlu-sion of the anterior intracranial circulation affecting thedistal intracranial internal carotid artery (ICA) and/orproximal middle cerebral artery (MCA) and/or the proxi-mal anterior cerebral artery (ACA). Many of these chil-dren who develop severe intracranial stenosis alsodevelop moyamoya phenomenon. The stenoses arelocated in sites that can be readily evaluated by TCD.Based on the original data5,6 demonstrating that TCDcould identify children at high risk of stroke, the StrokePrevention in Sickle Cell Disease (STOP) trial was under-taken.1 In the STOP trial, 130 children identified by TCDas being at high risk of stroke were randomized to receiveeither transfusion or standard care. These 130 childrenwere entered into the trial based on their having 2 TCDexaminations that demonstrated time-averaged mean ofthe maximum (TAMM) velocities of ≥200 cm/s in one ormore of the MCAs or terminal ICAs. Over an average

354 Copyright © 2001 by the American Society of Neuroimaging

Received March 2, 2001, and in revised form May 2, 2001.Accepted for publication May 3, 2001.

From the Department of Neurology, Medical College ofGeorgia, Augusta.

Address correspondence to Dr Nichols, Department ofNeurology/BP 3147, Medical College of Georgia,Augusta, GA 30912. E-mail: fnichols@neuro.mcg.edu.

follow-up of about 20 months, there was 1 stroke in the 63children randomized to transfusion and 11 strokes in the67 randomized to standard care, which represented agreater than 90% reduction in stroke incidence in thetransfusion population. As a result of the findings of thisstudy, the National Institutes of Health released a clinicalalert on September 18, 1997, which stated:

The STOP Trial confirmed that TCD can identify chil-dren with sickle cell anemia at high risk for first timestroke. Since the greatest risk of stroke occurs in earlychildhood, it is recommended that children ages 2-16receive TCD screening. Screening should be conductedat a site where clinicians have been trained to provideTCDs of comparable quality and information content tothose used in the STOP trial and to read them in a mannerconsistent with what was done in STOP. . . . It is recom-mended that centers that wish to start screening childrenwith sickle cell anemia for stroke risk do studies to com-pare their current equipment with STOP trial TCD equip-ment. . . . Although the optimal timing is not known, re-screening should occur approximately every 6 months.

The TCD velocity criteria used in STOP for classificationof stroke risk (Table 1) apply only to children with sicklecell anemia who have not had a previous stroke. Althoughthe clinical alert suggests rescreening every 6 months, wethink that those children with normal velocities can berescreened once a year, those with conditional velocitiesshould be rescreened within 3 to 6 months, and those withan abnormal velocities should undergo repeat screeningwithin the next few weeks (and if the second study is alsoabnormal, should be offered transfusion therapy at thattime).

STOP TCD Examination

The STOP TCD protocol is one that requires attention todetail. It is a focused examination designed to obtain thehighest TAMM. The STOP TCD scanning protocol usesan absolute cut point of a TAMM ≥ 200 cm/s in the distalICA or MCA to determine stroke risk. It became clearearly in the study that to obtain the velocities used in thisstudy, the TCD sonographers had to perform meticulousexaminations, expending extra effort to identify the high-est velocities rather than simply obtaining a Dopplerwaveform and moving to the next depth.

For the STOP findings to be applied correctly to deter-mine stroke risk in children with sickle cell disease, theTCD interpreters must use the TAMM velocity and notother velocities.

TCD users who wish to use the STOP data to interpretTCD examinations performed on sickle cell patientsshould always remember that although most Dopplerarterial examinations use peak systolic velocity to deter-mine severity of stenosis, the STOP TCD criteria are

based on the TAMM velocity. Children with sickle celldisease have higher velocity flows than other children,and significantly higher velocity flows than adults. A childwith sickle cell disease who is at low risk of stroke has aTAMM velocity of about 130 cm/s. A peak systolic veloc-ity of 200 cm/s is normal for a child with sickle cell dis-ease; however, a TAMM velocity of 200 cm/s is severelyabnormal.

Brief Review of the TAMM

The use of the TAMM was initially proposed by Aaslid. Inthe first edition of his book on TCD,7 Aaslid discussed thetopic of choice of velocities:

The most practical Doppler reading of velocity is that ofthe spectral envelope or outline. This parameter is rela-tively easy to define even if the signal is barely above theinstrument noise. Furthermore, if flow in curved segmentsis observed, the maximum Doppler shift will correspondto the part of the segment closest to being “in line” withthe ultrasonic beam. The outline reading thus has aninherent error minimization when measuring absoluteflow velocities.7(p28)

Aaslid then noted,

Traditionally, systolic, diastolic, and mean values are usedto describe pressure, flow and velocity in the arterial sys-tem. Of these values, the mean carries the highest physio-logical significance because it depends less on central car-diovascular factors such as heart rate, contractility, totalperipheral resistance and aortic compliance than do sys-tolic or diastolic values. Moreover, the time-mean (of themaximum Doppler shift) velocity correlates better withperfusion than the peak and trough values.7(pp56-57)

In the second edition of Aaslid’s book,8 it was noted that

Nichols et al: STOP Study Guidelines for TCD Testing 355

Table 1. Stroke Prevention in Sickle Cell Disease (STOP) Clas-sification of Transcranial Doppler Results in ChildrenWith Sickle Cell Anemia

Normal TAMM <170 cm/sConditional TAMM >170 but < 200 cm/s in the middle

cerebral artery and/or distal internal carotidartery

TAMM >170 in posterior cerebral artery oranterior cerebral artery

Abnormal TAMM ≥200 cm/s in the middle cerebralartery and/or terminal internal carotid artery

Inadequatea

TAMM = time-averaged mean of the maximum.a. Study Was Unable to Be Read.

the mean flow velocity utilized in Transcranial Doppler isreferred to as a time mean of the peak velocity envelope,the envelope being a trace of the peak flow velocities as afunction of time.8(p14)

Because almost all TCD reports used TAMM velocity, theTAMM velocity was used in the development of theSTOP TCD velocity criteria (Figs 1, 2).

356 Journal of Neuroimaging Vol 11 No 4 October 2001

Fig 1. Visual measurement of the time-averaged mean of the maximum.A spectral outline of the highest velocities of the wave-form is shown.The horizontal cursor is placed so that the area above the line and under the peak of the waveform outline (A) is thesame as the area below the line and above the waveform outline (B).

Fig 2. Actual spectral trace of Figure 1 demonstrating placement of the cursor on the waveform. The gain has been set so thatthe waveform follower/envelope is accurately tracking the maximum velocity. On this well-defined waveform, the visual measure-ment of time-averaged mean of the maximum (TAMM) and the computer measurement of the TAMM are the same, 70 cm/s. Thecursor measurement is in the second box from the top on the right side of the screen. The computer-measured TAMM is in thefourth box from the top on the right.

Other Mean Velocities

There has been a great deal of confusion about the termmean velocity. On the original TCD equipment, theTAMM was reported simply as the mean. As a result ofthis, much of the original TCD literature simply used theterm mean and did not specify that it was the TAMM.Over the past few years, TCDI has come into use. TCDIcan calculate a number of mean velocities that are not theTAMM. None of these other calculated means are theTAMM and, therefore, cannot be applied to the STOPdata set.

On some TCDI units, a vertical line can be brought upon the screen and used to measure the highest velocity at aspecific point. When this measurement is engaged, at thesame time that the highest velocity is measured the com-puter can calculate the mean velocity by averaging all ofthe velocities in the spectral profile at that point in time.This measurement is sometimes referred to as the instan-taneous mean (see Fig 3).

The computer can summate a series of instantaneousmeans obtained over the time course of one or morewaveforms and develop a time average of the mean of themean velocities (see Fig 4). Some ultrasound units identifythis as the time-averaged mean, so it can potentially beconfused with the TAMM. This time-averaged mean ofthe mean velocity is not the same as the time-averagedmean of the maximum velocity. If the sonographerreviews the image, the mean velocity line that is drawnwill track within the body of the waveform and will notfollow the peak velocities. This time-averaged mean ofthe mean velocity is significantly lower than the TAMM.

Yet another mean is the intensity weighted mean acrosstime. This represents the mean of every point of the fastFourier transform weighted by the intensity of thereflected signal above (or below) the zero line. This veloc-ity is significantly lower than the TAMM.

Many centers have calculated the mean using a for-mula that adds one third of the peak systolic velocity plustwo thirds of the end diastolic velocity. This formula virtu-ally always results in velocities that are slightly lower thanthe TAMM. In a preliminary study of 20 sickle cellpatients, we found that this calculation typically resultedin velocities that were 4% to 10% lower than the TAMM.In some instances, the formula-calculated mean was actu-ally higher than the measured TAMM. We did not pursuethe reasons for this difference, but it appeared that thepulsatility index and heart rate affected this calculation.Because this calculation does not reliably give the sameTAMM velocity as used in STOP, this calculation shouldnot be used to determine the TAMM.

Nichols et al: STOP Study Guidelines for TCD Testing 357

Fig 3. (Top) Instantaneous mean. The vertical cursor isplaced so that it measures the peak velocity. This is displayedon the right side of the screen as 0.57 m/s. Under this numberis a mean of 0.20.m/s. This mean represents an average of allthe velocities in the waveform that fall along this vertical cursor.(Bottom). Same waveform. The vertical cursor has beenmoved very slightly to the right of the position in the top image.The peak velocity is still measured as 0.57 m/s, but the meanat this point in time is 0.39 m/s. It can be seen that this mean isnot only the same as the time-averaged mean of the maximumbut also very sensitive to cursor placement.

On the Nicolet TC-2000 TCD unit, which was used inthe STOP study, there is a waveform follower (envelope)that tracks the highest velocities of the velocity profile.The computer then averages the highest velocities overtime and determines the TAMM. Transcranial colorDoppler imaging (TCDI) units use different terms for thesame velocity: Acuson calls its highest velocity profiletrace the TAMx (time average of the maximum) velocity;ATL calls its version of the TAMM the TAP (time averageof the peak) velocity.

Angle of Incidence

The Nicolet TC-2000 TCD unit uses dedicated (or“blind,” since there is no B-mode image data) Doppler.Because the artery cannot be visualized, angle correctioncannot be applied. The TCD unit assumes that the kHzshift is obtained at an optimal angle of insonation (0° or180°) and that the angle of correction is not needed to cal-culate the velocity. Because all of the STOP velocity datawere developed using blind Doppler, angle correctionshould not be used if an examination is performed usingTCDI, since angle correction may potentially result in cal-culated velocities that are higher than those that wouldhave been obtained using the STOP protocol. Our recentexperience comparing TCDI with STOP TCD has beenrecently published.9

STOP TCD Scanning Protocol

Children have smaller head diameters than adults. Vesselidentification by TCD is based on expected depth of arte-rial segments, direction of flow, and spatial relationshipsto other identified vessels. Calipers are used to measurehead diameters as the distance between the posterior as-pects of one transtemporal window to the posterior aspectof the other transtemporal windows (the calipers are

placed so that the caliper tips lie just anterior to each ear,just above the zygomatic arch). Nomograms for differenthead diameters were developed to provide sonographerswith data on expected depths for different landmarks suchas the ICA bifurcation and the top of the basilar (Table 2).

Children have thinner skulls than adults and largetranstemporal acoustic windows. Adequate power to pen-etrate the skull is usually easily achieved. The large win-dows allow the sonographer to manipulate the probe tooptimize alignment with the artery to obtain the highestvelocity.

Children should be scanned only when clinically sta-ble. Hypoxia, hypercarbia, fever, hypoglycemia, andworsened anemia can all increase cerebral blood flow(CBF) and flow velocity. Sickle chest syndrome, pneumo-nia, splenic sequestration, hemolytic crisis, hypogly-cemia, and potentially other processes can all result in anincrease in CBF, and TCD velocities, over the patient’sbaseline. Hypocarbia and recent transfusions candecrease CBF and flow velocity. Because of the potentialimpact of alterations of any of these variables on TCDvelocities, the results of TCD scans performed when chil-dren with sickle cell are admitted to the hospital for medi-cal illnesses should generally not be used to determinestroke risk.

358 Journal of Neuroimaging Vol 11 No 4 October 2001

Fig 4. This image demonstrates the difference between the time-averaged mean of the maximum (TAMM) and the mean of allvelocities. On this ultrasound unit, the TAMM is referred to as the time average of the peak (TAP) velocities. The TAP is calculatedusing the waveform follower and tracks the highest velocities.The time average of the mean (TAM) averages all of the velocities ateach point in time. It is tracked by the black line that lies within the waveform. As can be seen, the TAP is significantly higher (158cm/s) than the time average of the mean (85.9 cm/s).

Children should lie quietly but should not be allowedto go to sleep because the increase in CO2 with sleep canelevate CBF and velocities. Because of the large numberof variables that can potentially affect flow velocity, STOPrequired that a child have 2 abnormal TCDs separated byat least 2 weeks before the child was considered definitelyabnormal and could be randomized in this study.

The STOP TCD scanning protocol requires the exam-iner to meticulously track the arterial segments and opti-mize the signal at each depth to obtain the highest velocitypossible. Because the velocity risk profile for stroke devel-opment is based on the highest velocity in the MCA andterminal ICA, the major focus is on identifying the highestvelocity in these areas. The MCA should be identified andtracked to as shallow a depth as possible, usually <40 mm.The artery is then tracked in 2 mm depth increments, withthe signal optimized and recorded at each depth. The ICAbifurcation is identified, and recordings are made of theACA and terminal ICA ≥4 mm deeper than the ICAbifurcation. The posterior cerebral arteries are then iden-tified, tracked as shallowly as possible, and tracked andrecorded in 2 mm depth increments to the top of the basi-lar at the midline.

Signal Optimization

Signal optimization consists of a combination of effortsused by the sonographer to obtain the highest velocity.The initial goal is to identify a transtemporal window withthe best angle of insonation. Children have large tempo-ral windows when compared to adults. Unlike manyadults who have only 1 window, children have severalwindows, and the sonographer has to identify which ofseveral windows is best. The sonographer searches for thewindow where the sharpest waveforms and highest veloc-ities can be obtained. Once the best window has beenidentified, the sonographer attempts to obtain the highestvelocity by making a series of minor probe manipulationsthat may involve either minor changes in angulation orsliding of the probe to obtain a better angle of insonation.

Because the highest TAMM predicts stroke risk, it iscrucial that the sonographer undertake this painstaking/meticulous search for the highest velocity at each depth.Failure to identify a high velocity may result in incorrectrisk prediction for a child. Even when a strong Dopplersignal is detected, minor manipulations of the probe willsometimes detect a much higher velocity signal that is“hiding” in the background. The sonographer must alsopay close attention to audible clues suggesting the pres-ence of local high velocities: these include turbulence,high-pitched pulsatile hissing in the background (which isnot displayed), and sudden cutoff of visual display of thewaveform. As the vessel is tracked, the sonographer mayhave to continue to make minor adjustments of the probeto identify the highest velocity at each depth.

The sonographers are instructed to obtain as clean andsharp a signal as possible, with the best possible signal-to-noise ratio. They are also instructed to obtain the highestvelocity, even if that signal is not as crisp and clear as alower velocity signal at the same depth.

Measuring the Velocity

For STOP, all velocities were read off-line using a visualtechnique.9 The Nicolet TC-2000 has a computer-basedalgorithm for tracking the maximum velocities and deter-mining the TAMM. Unfortunately, many TCD record-ings have less than optimal signal-to-noise ratios, whichcaused mistracking of the waveform follower, which inturn results in inaccurate computer-determined measure-ments (Fig 5). For this reason, we developed a standard-ized reading protocol that was rigorously tested and foundto deliver extremely reproducible measurements. TheNicolet TC-2000 allowed manipulation of the baselineand gain controls. It also provided a horizontal line (cur-sor) that could be displayed on the screen and positionedat any point on the waveform to help with the measure-ments. The gain settings and baselines are adjusted tostandardized settings prior to reading the velocities.

Nichols et al: STOP Study Guidelines for TCD Testing 359

Table 2. Stroke Prevention in Sickle Cell Disease (STOP) Bitemporal Head Diameter Nomogram: Expected Arterial Depths (in mm)for Different Head Diameters

TranstemporalHead Diameter MCA-1a MCA ICA Bifurcation ACA PCA TOB

12 cm 30-36 30-54 50-54 50-58 40-60 6013 cm 30-36 30-58 52-58 52-62 42-66 6514 cm 34-40 34-62 56-64 56-68 46-70 7015 cm 40-46 40-66 56-66 56-72 50-70 75

MCA = middle cerebral artery, ICA = internal carotid artery, ACA = anterior cerebral artery, PCA = posterior cerebral artery, TOB = top of the basi-lar. As a rough guide, the TOB lies at the midline and the ICA bifurcation is in general about 10 mm shallower than the TOB.a. MCA-1 is shallowest depth at which the MCA can be recorded.

The gain is adjusted as follows. The gain is increased tothe point that the waveform follower first begins to mis-identify background noise as signal; when this happens,the gain is then decreased to next highest level. If there iswraparound signal causing the waveform follower to mis-identify the wraparound as part of the signal, the baselineis adjusted to remove the wraparound. These adjustmentsof gain and baseline are performed by every reader priorto reading so that each reader reads velocities measured atthe same gain and baseline settings. This standardizationresults in highly reproducible velocity measurement. Whenthe visually guided readings using the STOP TCD read-ing protocol are compared to the computer-determinedTAMM on waveforms that allow optimal tracking of themaximal velocities by the waveform follower, the veloci-ties are usually within ≤5 cm/s of each other (see Fig 6).

In those TCD recordings that have excellent signal-to-noise ratio, the waveform follower will tightly track thehighest velocities and will accurately calculate theTAMM. Under situations in which there is poor signal-to-noise ratio, the waveform follower will not accuratelytrack the highest velocities and may overread orunderread the TAMM (see Fig 5). The STOP reading pro-tocol allows one to accurately read and record the veloci-ties in those recordings where the waveform is welldefined but where the signal-to-noise ratio is poor and thewaveform follower cannot accurately track the highestvelocities. This visually guided reading technique wasstandardized as follows. Imagine that the waveforms rep-

resent mountain peaks and valleys. Draw a line across thewaveforms so that if the peaks of the waveforms/moun-tains were pushed over, they would fill the valleys (seeFigs 1, 2, 6). This represents the TAMM line. This line typ-ically lies at the level of the “shoulder” of the waveform.However, this rule of thumb does not work as well withwaveforms with unusually high or low pulsatility. Thereader can rapidly calibrate his or her eye by workingwith sharply defined waveforms with good signal-to-noiseratio and comparing the visually read TAMM with thewaveform follower measured TAMM.

Conclusion

The STOP study demonstrated that TCD can be used toidentify a population of children with sickle cell diseasewho are at high risk of developing a stroke. The risk deter-mination is based on the TAMM and not on peak systolicor other velocities. To use the STOP TCD criteria, theTCD examinations and readings should be performed ina manner similar to those used in STOP. The TCD exami-nations performed in STOP were very focused, with theaim of identifying the highest velocity in the MCA anddistal ICA. The STOP TCD reading protocol resulted invelocities similar to those obtained by the computer onoptimal signal-to-noise ratio signals and allowed accuratemeasurement of velocities in those recordings wherethere was adequate signal outline to measure, but wherethe signal-to-noise ratio was poor and the waveform fol-lower could not track the waveform reliably. When the

360 Journal of Neuroimaging Vol 11 No 4 October 2001

Fig 5. An example in which the gain has been set too high so that there is a poor signal-to-noise ratio. The waveform follower isunable to accurately track the peak velocities, resulting in the computer overestimating the velocities.

Nichols et al: STOP Study Guidelines for TCD Testing 361

Fig 6. Cursor placement (refer to Fig 1 for definitions of A and B).This series demonstrates visual placement of the cursor. (Top)The horizontal cursor is much too low and the peaks are much too large for the valleys (A is much greater than B), and the resultingvelocity of 148 cm/s is less than the true time-averaged mean of the maximum (TAMM). (Middle) The cursor is placed appropri-ately so that A = B, and the measured velocity, 174 cm/s, is correct. (Bottom) The cursor is placed “too high” so that A is muchsmaller than B, and the resulting velocity measurement, 205 cm/s, is higher than the correctly measured TAMM.

TCD examinations and readings are performed accord-ing to STOP protocol on children with sickle cell disease,those children at high risk for development of stroke canbe identified and managed with transfusion to dramati-cally decrease their risk of stroke.

References

1. Adams RJ, McKie VC, Hsu L, et al. Prevention of first strokeby transfusions in children with sickle cell anemia andabnormal results on transcranial Doppler ultrasonography.N Engl J Med 1998;339:5–11.

2. Adams RJ, McKie VC, Brambilla D. Stroke prevention trialin sickle cell anemia (“STOP”): study design. J Control ClinTrials 1998;9:110–129.

3. Adams RJ, Brambilla D, Vichinsky E, et al. Stroke preven-tion trial in sickle cell anemia (STOP study): risk of stroke in

1933 children screened with transcranial Doppler (TCD).Stroke 1999;30:238.

4. Ohene-Frempong K, Weiner SJ, Sleeper LA, et al.Cerebrovascular accidents in sickle cell disease: rates andrisk factors. Blood 1998;91:288–294.

5. Adams R, McKie V, Nichols F, et al. The use of transcranialultrasonography to predict stroke in sickle cell disease. NEngl J Med 1992;326:605–610.

6. Adams RJ, McKie VC, Carol EM, et al. Long term risk ofstroke in children with sickle cell disease screened withtranscranial Doppler. Ann Neurol 1997;42:699–704.

7. Aaslid R, ed. Transcranial Doppler Sonography. New York, NY:Springer-Verlag; 1986.

8. Fujioka KA, Douville CM. Anatomy and freehand exami-nation techniques. In: Newell DW, Aaslid R, eds.Transcranial Doppler. New York, NY: Raven Press; 1992.

9. Jones AM, Seibert JJ, Nichols FT, et al. Comparison oftranscranial Doppler imaging (TCDI) and transcranialDoppler (TCD) in children with sickle-cell anemia. PediatrRadiol 2001;31:461–469.

362 Journal of Neuroimaging Vol 11 No 4 October 2001