Comparison of radiation treatment delivery for pancreaticcancer: Linac intensity-modulated radiotherapy versushelical tomotherapyRobert Taylor,1 Krisha Opfermann,1 Brian D Jones,1 Lacy E Terwilliger,1 Daniel G McDonald,1
Michael S Ashenafi,1 Elizabeth Garrett-Meyer2 and David T Marshall1
Departments of 1Radiation Oncology and 2Biostatistics and Epidemiology, Hollings Cancer Center, Medical University of South Carolina, Charleston,
South Carolina, USA
R Taylor PhD; K Opfermann MD; BD JonesBS; LE Terwilliger ART; DG McDonald MS;
MS Ashenafi MS; E Garrett-Meyer PhD;
DT Marshall MD, MS.
CorrespondenceDr David T Marshall, Department of Radiation
that there are no actual or potential conflicts of
interest related to their production of this
manuscript. (This is based on the definition as
provided by the International Committee of
Medical Journal Editors.)
Disclaimers: None.
Submitted 13 June 2011; accepted 30 October
2011.
doi:10.1111/j.1754-9485.2012.02373.x
Abstract
Introduction: Intensity-modulated radiotherapy (IMRT) has been shown toreduce dose to organs at risk (OAR) while adequately treating tumour volume.This study quantitatively compares the dosimetric differences from step-and-shoot IMRT compared with helical tomotherapy (HT) for pancreatic headcancer.Methods: Twelve consecutive patients with non-metastatic, stage T3 or T4,unresectable pancreatic head cancer were planned for step-and-shoot IMRTas well as HT. Radiotherapy was planned to deliver 45.9 Gy to the clinicaltarget volume in 30 fractions with an integrated boost to 54 Gy to the grosstumour volume (planning target volume 5400 including a 1-cm set-upmargin). The uniformity index (UI) and conformity index (CI) were used tocompare the quality of target coverage, while the quality index (QI) comparedthe dosimetric performance for OAR.Results: Both methods were effective at covering the tumour with no signifi-cant difference in UI or CI. However, HT dosimetry exhibited superior sparingof OAR with significantly less stomach (mean QIStomV30 = 0.84, P = 0.006) andsmall bowel dosing (mean small bowel QISBV30 = 0.84, P = 0.005). HT reduceddose to the kidney receiving the highest dose but the overall volume of kidneyreceiving 18 Gy was not significantly different between the two systems,indicating that HT spread the dose more uniformly through the kidneys.Conclusions: Target coverage is equivalent between the two systems;however, HT shows significantly better sparing of the stomach and smallbowel. The decreased dose to OAR with HT is likely to improve the therapeuticratio in the radiotherapy of pancreatic head cancers.
Adenocarcinoma of the exocrine pancreas has a high rateof incidence within both the male and female populationswith a 5-year survival rate estimated at <5%.1 Whilesurgical resection is considered the most critical compo-nent of potentially curative treatment, radiotherapy (RT)is an important piece of the multimodality treatment forboth curative and palliative care.2,3
RT treatment of pancreatic cancer is complicated bythe close proximity of critical and sensitive structures(e.g. small bowel, stomach, kidneys and liver). More-over, pancreatic head cancer is generally treated with aprescription dose greater than the tolerance of the adja-cent organs at risk (OAR). However, progress in RTtechniques has improved the ability to spare adjacentOAR while delivering a therapeutic dose to the target.Over the last 20 years, improvements in RT treatment
bs_bs_banner
Journal of Medical Imaging and Radiation Oncology 56 (2012) 332–337
delivery have evolved from conventional to three-dimensional conformal RT and most recently tointensity-modulated RT (IMRT). IMRT in particularhas demonstrated superior ability of providing sharpdose gradients at the junction of target and adjacentnormal tissue4–6 in pancreatic and upper abdominalmalignancies.
Helical tomotherapy (HT) is a novel method of deliv-ering IMRT. It has several unique characteristics com-pared with the linac-based step-and-shoot technique.Here, radiation is delivered through 51 projections perrotation while the patient is continuously advancedresulting in a helical treatment pattern. A multileaf col-limator is used to shape the dose distribution at eachprojection and is operated in binary rather than thecontinuous mode more common to step-and-shootIMRT. The large number of beam projections results in amultitude of free parameters which potentially enables amore optimum target dose distribution while sparingadjacent normal tissue. HT has been found to providejust such a targeting advantage over step-and-shootIMRT in treatment of cancers such as head-and-neckcancer,7–9 craniospinal tumours,10 left breast cancer11
and retroperitoneal sarcoma.12 However, no comparisonof HT versus step-and-shoot IMRT for the treatmentof pancreatic head cancer has been done to date. Theprimary goal of this study was to investigate whetherHT plans are able to reduce dose to the OAR whileadequately treating the tumour volume as comparedwith step-and-shoot IMRT for pancreatic head cancer.
Methods
Patients
Twelve consecutive patients with cancer of the pancre-atic head were selected to participate in this planningstudy from among those enrolled in an InstitutionalReview Board-approved phase II protocol of neoadjuvantgemcitabine/oxaliplatin and cetuximab chemotherapyfollowed by RT with concurrent capecitabine. All patientsenrolled on trial had non-metastatic pancreatic cancerstage T3 or T4 with unresectable or borderline unresec-table disease. Every patient was determined to have
adequate functioning of both kidneys. All 12 patients hadalready completed their treatment at the time of thisdosimetric evaluation.
RT planning
RT was planned to deliver 45.9 Gy at 1.53 Gy per frac-tion to the elective nodal regions (planning targetvolume (PTV)4590) while simultaneously delivering54 Gy at 1.8 Gy per fraction to gross disease (PTV5400).The clinical target volume (CTV)4590 included grossdisease plus a 1-cm margin and the elective nodalregions, while the CTV5400 included only the grossdisease and a 1-cm margin. Patients were set-up supinewith a wingboard, Vac-Lok bag and/or Combifix immo-bilisation devices (CIVCO, Kalona, IA, USA). Four-dimensional (4D) free breathing CT simulation enabledquantification of target movement. Maximum targetexcursion was determined in the left–right, superior–inferior and anterior–posterior dimensions using 4D CT.The maximum excursion was then used to expand theCTV to patient-specific internal target volume (ITV);thereafter, a 0.5-cm expansion for set-up error in alldimensions was used to generate a PTV from the ITV.
Inverse plans were then developed for linac-basedstep-and-shoot IMRT and HT based on the same set ofplanning objectives and constraints as seen in Table 1.All volumes of interest were contoured in the ADACPinnacle(3)® system (Phillips Healthcare, Andover, MA,USA). For HT planning, the image sets with contourswere imported into the TomoTherapy® HI-ART treatmentplanning system (version 3.1.4.7; Tomotherapy Inc.,Madison, WI, USA). Linac-based IMRT plans weredevised in ADAC Pinnacle(3)® (Phillips Healthcare) withhomogeneity correction selected. Planning was based on16-MV photons and non-coplanar beams were used foroptimisation (direct aperture optimisation utilised). ForHT planning, the field width was set at 2.5 cm, modula-tion factor at 2.6 and the grid at 0.268 cm. The pitch,which is the distance travelled by the couch during onecomplete rotation divided by the field width, was set to0.43. HT utilises 6-MV photons. The primary aim was toprovide the most optimal plan that met treatment objec-tives while minimising dose to the OAR. Dosimetric data
Table 1. Treatment planning objectives used for both helical tomotherapy and step-and-shoot intensity-modulated radiotherapy plans
Liver Volume of liver less any CTV V20Gy < 67% V30Gy � 40% NA
Kidneys Contoured separately and expanded by 0.5 cm Hot kidney V18Gy � 75% Cold kidney V18Gy � 25% NA
Small bowel Entire small bowel volume as a compartment Dmean � 13 Gy V30Gy � 20% V45Gy � 10%
Stomach Volume of stomach less any CTV V30Gy � 50% V45Gy � 10% NA
CTV, clinical target volume; Dmean, mean dose; NA, not applicable; PTV, planning target volume; Vx%, volume receiving at least x% of the prescribed dose;
were evaluated based on predetermined criteria, and theIMRT plans were compared with the HT plans.
Metric models
In order to assess the quality of target coverage, theuniformity index (UI) and conformity index (CI) wereused. The UI is defined as follows:
UI = DD
5
95
where D5 and D95 are the minimum doses delivered to thehottest 5% and 95% of the PTV as previously described byWang et al.13 UIs that are closer to one are consideredmore homogenous in the dose delivery, which is a desir-able dosimetry endpoint. The UI measures the unifor-mity within the target region, while the CI measures theamount of the prescribed dose that is in the targetvolume. We have found that the most suitable assess-ment of CI, as previously described by Paddick,14 is
CITV
TV PIVPIV= ( )×
2
where TVPIV is the volume of the target covered by theprescription isodose, TV is the target volume and PIVis the prescription isodose volume. This CI formula hasthe advantage of accounting for both overtreatment andundertreatment. The CI ranges from zero to one and ascore of one indicates perfect conformity with the PIVexactly overlapping only the TV. The CI is important toconsider as it will characterise dose to all normal tissueand is not limited specifically to the contoured OAR.
To evaluate the dose to critical organs, a quality index(QI) for OAR is defined as follows:
QI = VV
doseTomo
doseLinac
where VdoseTomo is the volume of the OAR receiving a speci-
fied dose in the HT plan and VdoseLinac is the volume of the
OAR receiving the same specified dose in the linac plan.
Statistical analysis
The UI, CI and QI described above were created sepa-rately for each patient. The Wilcoxon signed-rank testfor matched pairs was used to calculate the two-sidedP-value. A two-tailed value of P < 0.05 was defined ashaving statistical significance. SPSS version 13.0 (IBMSPSS, Armonk, NY, USA) was used for the calculations.As the QI values are matched pair ratios, the distributionis of the lognormal type, so the logarithm of the QIvalues was used for calculations of statistical signifi-cance.
Results
Target coverage
Figure 1 illustrates typical isodose contours for bothplanning systems. Both HT and step-and-shoot IMRTplans appropriately treated the PTV in accordance withthe planning goal of 95% or greater of the prescribeddose delivered to 100% of the PTV as can be seen bycomparing the dose-volume histograms in Figure 2.The UI for the PTV5400 was not statistically differentbetween the two plans, with each performing nearlyidentically as seen in Table 2 (HT mean = 1.03, linacIMRT mean = 1.03; P = 0.695). The CI was also verysimilar between the two plans with no statistical differ-ence seen and no clear trend (P = 0.859).
OAR sparing
Both HT and step-and-shoot plans had no violationsof planning objectives for kidneys or liver. Per dosimetry
Fig. 1. Comparison of dose distribution between the linac-based step-and-shoot intensity-modulated radiotherapy plan (a) and helical tomotherapy plan (b) for the
same patient. The computed tomography slice is the same for each image. Thick red line = planning target volume (PTV)4590, thick blue line = PTV5400, thin
maroon line = 45.9-Gy isodose line, yellow line = 30-Gy isodose line and thin blue line = 18-Gy isodose line.
objectives, the kidney dose was limited in patients withfully functioning kidneys to 18 Gy in �75% in one kidneyand a volume of �25% in the other. All patients in bothHT and step-and-shoot plans were able to meet theseobjectives. The HT plan did succeed in significantlyreducing the dose to the kidney receiving the highest
dose (mean QIKidHotV18 = 0.83, P = 0.023) as seen inTable 3. However, this did not result in a significantdecrease to the overall kidney dose as the mean V18Gyfor total kidney volume (both kidneys combined) was30.6% for HT and 32.1% for IMRT (P = 0.209), indicatingthat the dose was redistributed more uniformly betweeneach kidney in the HT plan. Neither the liver volumereceiving at least 30 Gy nor the liver volume receiving atleast 20 Gy was significantly different between the twoplans as seen in Table 3.
The small bowel and stomach objectives were difficultto achieve in both planning systems due to anatomicalproximity and the dose limitations of the organs. TheIMRT plans had eight out of 12 and the HT plans hadthree out of 12 patients, which violated the small boweldosimetry objective of �20% of the volume receiving30 Gy or more. The HT plan was significantly better,however, in minimising this dose (mean QISBV30 = 0.84,P = 0.005). This trend was seen in the small bowelvolume receiving 45 Gy or more as well but was notstatistically significant. The dose to the stomach showeda similar pattern with the HT plan performing signifi-cantly better in the volume receiving 30 Gy or more(mean QIStomV30 = 0.84, P = 0.006), while the volumereceiving 45 Gy trended in favour of the HT plan but wasnot statistically significant.
Discussion
The uniformity and conformity are important for tumourcontrol and minimisation of dose to adjacent normaltissue. We found that both HT and step-and-shoot linac-based IMRT adequately treated tumour volume, and theuniformity and conformity were nearly identical betweenthe two approaches. The equivalence may be due to theprimary emphasis placed by the dosimetrist on adequatetreatment of the tumour. Additionally, the linac-basedIMRT plan has the ability to use non-coplanar beams andin our patients every linac-based plan included non-coplanar beams. Non-coplanar beams are not an optionwith HT due to the limitations of an enclosed couch. Thisincreased the degrees of freedom with the step-and-shoot IMRT plan and may have offset the advantage ofincreased number of beamlets enjoyed by the HT systemfor meeting the primary tumour control objectives.
It is useful to think of the degrees of freedom enjoyedby a system as capital which can be expended to achieveobjectives. After spending capital to achieve adequatetumour treatment, the leftover capital can be spent tospare the OAR. Two of the more important OAR are thekidneys. For all patients, the primary renal planningobjectives were met. However, the HT system was ableto minimise the dose to the kidney receiving the highestdose, but this gain was at the expense of dose to theother kidney as seen in Figure 2. The overall renalvolume receiving 18 Gy or more was nearly identical,suggesting the HT system redistributed dose more uni-
Fig. 2. Cumulative dose-volume histogram of the linac-based step-and-shoot
intensity-modulated radiotherapy plan (a) and helical tomotherapy (HT) plan (b)
for the same representative patient. The maximum dose to either kidney is
reduced by the HT plan, although the total volume of combined kidneys receiv-
ing �18 Gy is equivalent. The other organs are either equivalent or the HT
has reduced dose. The dark blue line = planning target volume (PTV)5400,
red = PTV4590, light blue = spinal cord, pink = cold kidney, dark orange = hot
kidney, maroon = large bowel, green = small bowel, purple = stomach and
light orange = liver.
Table 2. Results of uniformity index (UI) and conformity index (CI) analyses
formly to normal tissue through its helical geometry. Theother OAR doses were either equivalent or the dosingwas better with HT as seen in the dose-volume histo-grams. This implies that after achieving primary tumourcontrol, the HT system enjoyed more planning capitaland was able to create a better plan. This decreaseddose to the small bowel and stomach would be expectedto decrease anorexia, nausea and vomiting, some of themost common acute and near-acute toxicities seen withRT to the upper abdomen.
In any comparison of different planning and deliverysystems, there are several factors that make a strictlyequivalent comparison difficult. The quality of the plandepends on the skill of the dosimetrist, the efficiency ofthe planning optimisation algorithm and the plan objec-tives. Two dosimetrists worked together on the plansto achieve the same set of objectives. All HT plans weredone by one dosimetrist and all linac-based IMRT planswere done by the other. Each dosimetrist was well expe-rienced in the respective planning system and workeduntil all objectives were accomplished and an optimalplan achieved. The plans were all reviewed and approvedby one physician (DTM). However, the overall inherentlimitations of this type of comparison limit our conclu-sions. Another difference in the two sets of plans wasthat the linac-based IMRT plans used 16-MV photons,which may not be available at some centres, while theHT plans used 6-MV photons. Sixteen-megavolt photonswere used in the linac-based IMRT plans because theyproduced better plans than 6-MV photons.
In conclusion, we compared the dosimetry of 12 pan-creatic head cancer patients for non-coplanar step-and-shoot linac-based IMRT with HT. The quality of the targetcoverage in terms of uniformity and conformity wasequivalent between the two systems. A significant reduc-tion in the kidney receiving the highest dose was seenwith the HT system, but the total kidney volume receiv-ing 18 Gy or more was equivalent. Liver dosing wasequivalent between the two treatment modalities. Sig-nificant improvements were seen in small bowel andstomach dosing with the HT system. The therapeuticratio of RT may be improved by increasing tumourcontrol or by decreasing toxicity. HT appears likely to
improve the therapeutic ratio of RT for pancreatic headcancers compared with step-and-shoot IMRT by decreas-ing dose to sensitive OAR that surround these cancers,namely, the small bowel and stomach.
References
1. Jemal A, Siegel R, Xu J, Ward E. Cancer statistics,2010. CA Cancer J Clin 2010; 60: 277–300.
2. Regine WF, Winter KA, Abrams RA et al. Fluorouracilvs gemcitabine chemotherapy before and afterfluorouracil-based chemoradiation following resectionof pancreatic adenocarcinoma: a randomizedcontrolled trial. JAMA 2008; 299 (9): 1019–26.
3. Rich T, Harris J, Abrams R et al. Phase II study ofexternal irradiation and weekly paclitaxel fornonmetastatic, unresectable pancreatic cancer:RTOG-98-12. Am J Clin Oncol 2004; 27 (1):51–6.
4. Van der Geld YG, van Triest B, Verbakel WF et al.Evaluation of four-dimensional computedtomography-based intensity-modulated andrespiratory-gated radiotherapy techniques forpancreatic carcinoma. Int J Radiat Oncol Biol Phys2008; 72 (4): 1215–20.
5. Kataria T, Rawat S, Sinha SN et al. Intensitymodulated radiotherapy in abdominal malignancies:our experience in reducing the dose to normalstructures as compared to the gross tumor. J CancerRes Ther 2006; 2 (4): 161–5.
6. Brown MW, Ning H, Arora B et al. A dosimetricanalysis of dose escalation using two intensitymodulated radiation therapy techniques in locallyadvanced pancreatic carcinoma. Int J Radiat OncolBiol Phys 2006; 65 (1): 274–83.
7. van Vulpen M, Field C, Raaijmakers CP et al.Comparing step-and-shoot IMRT with dynamichelical tomotherapy IMRT plans for head-and-neckcancer. Int J Radiat Oncol Biol Phys 2005; 62 (5):1535–9.
8. Sheng K, Molloy JA, Read PW. Intensity-modulatedradiation therapy (IMRT) dosimetry of the head andneck: a comparison of treatment plans using linearaccelerator-based IMRT and helical tomotherapy. IntJ Radiat Oncol Biol Phys 2006; 65 (3): 917–23.
Table 3. Quality indices (QIs) for organs at risk
Description Symbol Mean SD Min Max P
QI of the volume of the hottest kidney receiving 18 Gy or more QIKidHotV18 0.83 0.19 0.59 1.25 0.012
QI of the volume of the coldest kidney receiving 18 Gy or more QIKidColdV18 1.51 1.14 0.91 5.00 0.071
QI of the volume of the liver receiving 30 Gy or more QILiverV30 0.99 0.23 0.70 1.53 0.388
QI of the volume of the liver receiving 20 Gy or more QILiverV20 1.01 0.16 0.84 1.43 0.695
QI of the volume of the stomach receiving 45 Gy or more QIStomV45 0.97 0.17 0.62 1.21 0.695
QI of the volume of the stomach receiving 30 Gy or more QIStomV30 0.84 0.13 0.64 1.09 0.006
QI of the volume of the small bowel receiving 45 Gy or more QISBV45 0.91 0.18 0.49 1.13 0.099
QI of the volume of the small bowel receiving 30 Gy or more QISBV30 0.84 0.15 0.48 1.09 0.005
9. Lee TF, Fang FM, Chao PJ et al. Dosimetriccomparisons of helical tomotherapy and step-and-shoot intensity-modulated radiotherapy innasopharyngeal carcinoma. Radiother Oncol 2008;89 (1): 89–96.
10. Mavroidis P, Ferreira BC, Shi C et al. Comparisonof the helical tomotherapy and MLC-based IMRTradiation modalities in treating brain andcranio-spinal tumors. Technol Cancer Res Treat2009; 8 (1): 3–14.
11. Goddu SM, Chaudhari S, Mamalui-Hunter M et al.Helical tomotherapy planning for left-sided breastcancer patients with positive lymph nodes:comparison to conventional multiport breast
technique. Int J Radiat Oncol Biol Phys 2009;73 (4): 1243–51.
12. Pezner RD, Liu A, Han C et al. Dosimetric comparisonof helical tomotherapy treatment and step-and-shoot intensity-modulated radiotherapy of retroperito-neal sarcoma. Radiother Oncol 2006; 81 (1): 81–7.
13. Wang X, Zhang X, Dong L et al. Effectiveness ofnoncoplanar IMRT planning using a parallelizedmultiresolution beam angle optimization method forparanasal sinus carcinoma. Int J Radiat Oncol BiolPhys 2005; 63: 594–601.
14. Paddick I. A simple scoring ratio to index theconformity of radiosurgical treatment plans.J Neurosurg 2000; 93 (Suppl 3): 219–22.
