Radiotherapy & Imaging
a report by
C l i f t o n D av i d F u l l e r , MD , and Todd J S c a r b ro u g h , MD
Department of Radiatioin Oncology, University of Texas and Professor of Radiation Medicine,
Oregon Health & Science University
Fiducial markers are emerging as a standard tool forimage-guided radiotherapy (IGRT). The markers areimplantable devices designed to act as reliable surrogatesfor imaging anatomic structures of interest. Fiducialmarker techniques were originally developed in the pre-conformal radiotherapy era for positional verification oftissues that were not easily visualized using portal X-rayfilm imaging for patient alignment. Soft tissue structuresthat were relatively mobile could be more readily seenwhen radio-opaque seeds or wires were implanted in ornear organs of interest. However, since comparativelylarge margins around target volumes were used, dosedelivery limitations overshadowed imaging accuracy. Inaddition, routine and efficient ‘online’ (at the time oftreatment) implementation of patient set-up error analysisand patient repositioning was problematic in the past.
Modern radiotherapy increasingly utilizes conformaldelivery techniques, which necessitate more accuratepatient positioning and tumor targeting.1 Brought aboutby mechanical and computing advances, the ability todeliver radiation doses to the tumor while sparing normaltissue using 3-D conformal radiotherapy (3D-CRT),intensity modulated radiotherapy (IMRT) or otherconformal treatments, such as proton therapy, allows thepotential for reduction of toxicity and—through carefuldose escalation—potentially better outcomes. It is hopedthat by keeping ‘tight margins’ around target volumesthese dual goals may be attained.
As a practical matter, improved dose delivery precisionnecessitates a higher degree of accuracy in targeting thetumor (target delineation) and aligning the patient so asto reproducibly confirm that the tumor is in the sameplace in 3-D space as the prescribed conformal dosedistribution throughout therapy (positional verification).However, since even small geometric deviations fromthe treatment plan may result in substantial changes indose distribution,2 margin reduction must be matchedwith IGRT techniques lest margin reduction result inpoorer clinical outcomes.3
While fiducial markers have been implemented in headand neck, gastrointestinal, hepatic, and other sites, the vastmajority of existing data has been generated in prostate
cancer and, for the purposes of this manuscript, this organsite will be exclusively focused on. Prostate cancerdemonstrates a dose–response profile such that increasingdelivery of dose to tumor is associated with betteroutcomes.However, prostate radiotherapy presents severalunique issues for which fiducial markers may be a usefultool.The prostate is difficult to image using standard X-ray films, lies in proximity to hollow organs (the bladderand rectum), which are at risk of radiotherapy-inducedtoxicity, and may be displaced due to filling of theseadjacent organs.As such, the prostate represents a ‘movingtarget’ as changes in bladder and bowel filling can displacethe prostate significantly between, or even during,radiotherapy fractions.4 Since the prostate is not attacheddirectly to bony anatomy, prostate motion differssubstantially from pelvic bony anatomic position.Addinglarger margins to the prostate increases dose to bladderand rectum, improving target coverage but also increasingthe probability of toxicity: this is a less than optimalsolution to ensure tumor coverage. The problem ismagnified when steep dose gradients such as with 3D-CRT or IMRT are employed or when dose escalation isdesired.5 Since gross tumor volume and microscopicextent of disease cannot be reduced, the most reasonablemechanism for reduction of margins is increased accuracyof targeting and patient alignment.6,7
For years, the concept of visualizing intra-prostaticimplanted seeds has been a logistical consideration in theearliest conformal radiotherapy techniques (implantbrachytherapy).8–12 Indeed, several series have implantedfiducial markers during brachytherapy with I125 seeds inexpectation of subsequent boost external beamradiotherapy.10 As the concept of implanted devices forexternal beam targeting has evolved, numerouspermutations on the fiducial marker concept have arisen.The most frequently utilized devices in most series havebeen gold seeds, making them suitably visualized using avariety of imaging methodologies.Typically, the seeds arecylindrical, allowing ease of insertion via a needle bytransrectal ultrasound, and have surface features thatprevent seed migration after insertion. Varying sizes ofseed marker, including custom lengths, may be ordered.Carbon fiducials hold the promise of visibility oncomputed tomography (CT) imaging and producing less
Fiduc ia l Marker s in Image-gu ided Rad io therapy o f the Pros ta te
U S O N C O L O G I C A L D I S E A S E 2 0 0 6 75
Todd J Scarbrough, MD, is MedicalDirector of the MIMA Cancer Center,Melbourne, FL, and Adjunct AssociateProfessor of Radiation Medicine,Oregon Health & Science University,Portland, OR.
Clifton David Fuller, MD, is aresident in the Department ofRadiation Oncology and Scholar inHuman Imaging/Radiobiology in theGraduate Division of RadiologicalSciences at the University of TexasHealth Science Center at San Antonio, TX. His researchcenters on imaging applications in radiotherapy.
Fuller_book.qxp 12/1/07 12:13 pm Page 75
artifact as compared with gold (Civco/NMPE,Lynnwood,WA). Strings of markers on a bio-absorbablestrand are available (Best Medical, Springfield,VA), as aremetallic coils (Visicoil, Radiomed, Tyngsboro, MA)13.Both strings and coils are designed to span the whole ofthe gland and thus potentially provide a readily imagedand positionally stable surrogate. Another novel conceptdeveloped recently is encapsulation of radiofrequencytransponders within implantable markers; the signalreflected by these implanted transponders can be used towirelessly track prostate motion in realtime duringtherapy without reliance on radiographic imaging.14,15
While at present no established data sets havecompared product usability across fiducial markersystems, a series from the University of Manchestercompared marker sizes ranging from 3mm to 8mm inlength, and 0.8mm to 1.1mm diameter, usingelectronic portal images, determining that markers<5mm in length or <0.9mm in diameter were poorlyvisualized.16 Whether this observation holds true for a
variety of image guidance techniques (such aselectronic portal imaging, cone-beam CT, orthogonalkilovoltage X-ray, ultrasound, and CT)17 remains to beascertained. As several authors have noted, the wideempiric utilization of image-guided and conformalradiotherapy techniques has, to an extent, outpaced theavailability of rigorous intra- and inter-platformcomparisons of technical and clinical parameters.
Accruing data regarding fiducial markers have been putforth within the radiotherapy literature. Seminal datafrom the University of Michigan used fiducial markerplacement in concert with orthogonal port films,demonstrating a maximum displacement of 7.5mm in asingle directional axis, with most offsets of 0–4mm.18,19
Similarly, Crook et al. found similar directional offsets,with posterior mean offsets of 5.6mm; and inferiordirectional shifts averaging 5.9mm.20 These early datasets also confirmed that prostate motion was notuniform and random, but exhibited a predominantdirectional component, related to rectal and bladderfilling. Until the widespread implementation of the tightdose margins possible with IMRT, utilization remainedlimited to a few academic institutions. However, theseearly experiences using port films paved the way forfuture series,21 and established the clinical feasibility offiducial marker placement.19 Since these initial efforts,many groups have sought to implement fiducial markersas part of processes designed to afford positionalverification as well as image co-registration.14,21–28
Early adopters of seed implantation demonstrated thatimplementation of fiducial markers in clinical practicecould be achieved with minimal incremental work flowadjustment.Over time, as with any alteration to treatmentplanning protocols, efficiency of use improves rapidly afteran initial learning curve.At the MIMA Cancer Center inMelbourne, FL, patients arrive for an appointment threeto five days before their treatment planning CT. Thepatient is situated in the left lateral decubitus position, andtransrectal ultrasonography is used to localize the prostategland. Four gold seeds are placed via preloaded needlesand a transrectal approach near the apex, mid-gland, andleft and right base positions within the prostate usingsagittal ultrasound guidance without any anesthesia. Atleast three days are then allowed to pass before CTsimulation to allow residual swelling to resolve. Afterplacement, the patient may be readily imaged bytreatment planning CT simulation, and the seeds can beobserved easily on digital reconstructed radiographs. Atthis time, if desired, the seeds may be delineated at thesame time as clinically relevant target volumes and organsat risk.The MIMA Cancer Center has recently initiatedthin-slice magnetic resonance imaging (MRI) scanning ofpatients about one week after marker placement. Usingcustom MRI sequences as outlined by Huisman et al.,29
Radiotherapy & Imaging
76 U S O N C O L O G I C A L D I S E A S E 2 0 0 6
Figure 1: Axial CT-scan Sowing Carbon Fiducial Marker(Left Side, with Fiducial Detail in Inset) and Gold SeedMarkers (Right)
Figure 2: Digital Reconstructed Radiograph ShowingGold Marker Seeds
Fuller_book.qxp 17/1/07 12:54 pm Page 76
the small 1 x 3mm gold markers can be visualized onMRI and allow perfect prostate MRI-to-CT fusion.Prostate anatomy target delineation is improved with useof MRI, and the problem of registration/fusion errors(dose distributions can only be planned on CT data sets)is overcome by marker-to-marker/MRI-to-CT fusion.
Much of the work using fiducial markers wasperformed in order to more accurately define physicalpoints within the prostate for imaging methodcomparison. In a proto-image-guidance paper, Sandleret al., in a series of 15 patients, all of whom receivedpelvic CT and retrograde urethrogram after transrectalultrasound using gold seeds implanted in the prostate,demonstrated the superiority of CT-based imaging forapex localization, as well as the potential for visualizingseeds on mega-voltage CT.30 Similarly, seed data wasused by Malone et al. to determine that barium enemasmight provide erroneous inputs for target localization.31
Since the most common alternate method in recent yearshas been ultrasound prostate positional verification,32 themajority of method comparison series recently havecompared seed and ultrasound techniques.Emerging datasuggest that ultrasound systematic error is high comparedwith seed localization imaging platforms. Langen et al.found ultrasound alignments to exhibit a systematicdifferential from the marker alignments in the supero-inferior and lateral axes, with random variabilityequivalent to or similar to the initial set-up, and significantinter-user variability. Van den Heuvel, using five to sixseeds for both ultrasound and electronic portal imaging(EPI) localization, showed significant improvement withultrasound was obtained in only the antero-posteriordimension, with no measurable improvement in twoother room axes.33 In contrast, seed localization wasrevealed to show significant improvement in set-upcapability.34 Serago et al. recently compared alignment toskin marks with fiducial marker localization by EPI,ultrasound, and kilovoltage X-ray imaging. Thiscomparison revealed ultrasound improves alignment toEPI-based positioning compared with skin marks alone,while kV fiducial marker imaging was most consistentwith EPI. They conclude by noting that bothmegavoltage and kilovoltage imaging have similaraccuracy profiles and both offer a substantial improvementover ultrasound.35
The authors’ own experience with ultrasound versusseed marker prostate localization represents the largestIGRT data set thus far presented in the literature. In thefirst report, it was concluded that margin reduction(versus ultrasound or non-ultrasound approaches) usingseed markers could be substantial. It was also concludedthat versus set-up to skin marks only, an ultrasoundIGRT approach offered no improvement or some slight
worsening of set-up accuracy. As part of theinvestigations into method comparison of prostateimaging techniques, data suggest that ultrasound andkVp approaches have poor agreement and should notbe utilized interchangeably.36 Additionally, ultrasoundimaging exhibited systematic error components thatwould necessitate comparatively larger margins to safelyimplement in comparison to kilovoltage seed imaging.
As previously noted, the benefits available from dailyprostate alignment are only conferred when radiationdose may be preferentially delivered to target volumeswhile sparing adjacent structures. Target volumestypically are conceptualized using the InternationalCommission on Radiation Units and Measurements(ICRU) nomenclature, which specifies radiographicallyevident tumor as gross tumor volume (GTV) and theclinically expected region of microscopic extension oftumor as the clinical target volume (CTV). Since thesevolumes must receive the prescribed dose in order toeffect tumor control, their volumes may not beminimized without dire consequences. Practicallyspeaking, therefore, these volumes cannot be reduced, allreduction in irradiated volumes is performed byreducing the planning target volume (PTV).The PTV,according to van Herk,37 should represent “the volume,defined in treatment room coordinates, to which aprescribed dose must be delivered in order to obtain aclinically acceptable and specified probability that the
Fiduc ia l Marker s in Image-gu ided Rad io therapy o f the Pros ta te
U S O N C O L O G I C A L D I S E A S E 2 0 0 6 77
Figure 3: Plot of 1,019 Daily Kilovoltage X-ray Image-guided Shifts in 40 Patients using Fiducial Seed Markers
The 50% and 95% iso-surfaces represent a volume containing half and 95% of all
3-D shifts.Table at top shows the mean displacement and 95% iso-surface tolerance
dimensions in each principal room axis (all values in mm).
Fuller_book.qxp 12/1/07 12:10 pm Page 77
prescribed dose is actually received by the CTV, whichhas an uncertain location.”This safety margin is designedto conceptually account for a multitude of factors thatmight affect the capacity to deliver dose accurately. Byreducing the degree of variability of daily positioningthrough seed markers, it is possible to reduce therequired margins and, thus, with a smaller PTV, sparemore normal tissue.
Using established ‘recipes’ for margin calculation, theauthors’ group has determined that uniform PTVmargins of approximately 3mm would be required toaccount for random and systematic error inherent acommercially available seed-marker IGRT system. Incontrast, a comparison ultrasound localization systemwould require margins of approximately 9mm, onlymarginally improved from the 10–13mm marginsrequired without image guidance.24 While this findingcannot necessarily be generalized to other fiducialmarker-based imaging platforms, it does suggest apossible mechanism to reduce clinical toxicity byreducing the volume irradiated to 100% of theprescribed dose to tumor.
Fiducial marker utilization is not without inherent costs,however.Pollack38 notes three potential drawbacks of seedlocalization for prostate radiotherapy: invasiveness,automated positional registration error, and seedmigration.23 However, in our experience, these have notprecluded utilization. Seed placement is necessarilyinvasive, but the short (less than three minute) transrectal
insertion, even without anesthesia, has been exceedinglywell tolerated by patients. System accuracy for severalvendors has been measured at sub-millimeter ranges, andseed migration, while ever a concern, has not, at least inthe authors’ data set, been a noticeable phenomenon.Themore practical shortcomings of this, as with any technicalmodification, are increased time and resource costs.Whilefiducial marker tracking systems are more expensive,upfront hardware and staff costs of device placement andutilization must be considered. Even so, with trainedpersonnel and implementation of resource savingtechniques for placement (on-site transrectal ultrasoundimplantation) or daily set-up protocol modification,39
many obstacles may be minimized or circumvented.
The novelty of image-guided techniques and themultiplicity of platforms have precluded a definitivedetermination of the clinical import of fiducialmarkers on clinical outcomes in radiotherapy.40
Preliminary data, however, are enticing. Out of a seriesof 152 patients treated using fiducial marker imageguidance at MIMA Cancer Center to a prescriptiondose of 81Gy, 98% of patients were observedexperiencing no rectal toxicity (Radiation TherapyOncology Group (RTOG) grade 0) with theremaining 2% experiencing no more than RTOGgrade II rectal toxicity. No cases of grade III or greaterrectal toxicity have been observed to date (as measuredby Common Terminology Criteria v3.0).
While in the absence of long-term prospectiveoutcome data the true clinical value of fiducial marker-based systems remains to be determined, somepreliminary observations can be made. First, there is asubstantial amount of data that demonstrates statisticallysignificant margin reduction ability of fiducial marker-based prostate IGRT methods. Second, no other IGRTsystem or technique has definitively been shown to besuperior to fiducial markers in regards to marginreduction. Lastly, the authors’ own preliminary dataindicates a favorable toxicity profile using fiducialmarker-based prostate IGRT. In the opinion of theauthors, fiducial marker-based IGRT represents thestate of the art in conformal prostate radiotherapy andwill become (if not already) the standard of radiationmedicine care in this disease site.
Con c l u s i o n
Commercially available fiducial marker target localizationand positional verification techniques offer an attractivepremise for efforts for radiotherapy margin reduction inprostate cancer. While the technical and dosimetricaspects of fiducial marker co-registration are attractive,future study is required to determine whether increasedpatient positioning accuracy may be translated intoimproved outcomes and/or reduced toxicity. ■
A version of this article containing references can be found inthe Reference Section on the website supporting this briefing(www.touchoncologicaldisease.com).
Radiotherapy & Imaging
78 U S O N C O L O G I C A L D I S E A S E 2 0 0 6
Commercially available fiducial marker target localization and
positional verification techniques offer an attractive premise for
efforts for radiotherapy margin reduction in prostate cancer.
Fuller_book.qxp 12/1/07 12:10 pm Page 78
TGM2_ad.qxp 12/1/07 12:37 pm Page 79