dose escalation of chart in non-small cell lung cancer: is three-dimensional conformal radiation...

12
PII S0360-3016(99)00095-4 CLINICAL INVESTIGATION Lung DOSE ESCALATION OF CHART IN NON-SMALL CELL LUNG CANCER: IS THREE-DIMENSIONAL CONFORMAL RADIATION THERAPY REALLY NECESSARY? CAROL MCGIBNEY, F.R.C.S.I.,* OLA HOLMBERG, M.SC., ² BRENDAN MCCLEAN,PH.D., ² CHARLES WILLIAMS, F.R.C.R., PAMELA MCCREA, M. ECON.SC,PHIL SUTTON, D.C.R.(T), § AND JOHN ARMSTRONG, M.R.C.P.I., D.A.B.R.* Departments of *Radiation Oncology, ² Physics, Radiology, and § Therapeutic Radiography, St. Luke’s Hospital, Dublin, Ireland Purpose: To evaluate, preclinically, the potential for dose escalation of continuous, hyperfractionated, acceler- ated radiation therapy (CHART) for non small-cell lung cancer (NSCLC), we examined the strategy of omission of elective nodal irradiation with and without the application of three-dimensional conformal radiation technol- ogy (3DCRT). Methods and Materials: 2D, conventional therapy plans were designed according to the specifications of CHART for 18 patients with NSCLC (Stages Ib, IIb, IIIa, and IIIb). Further plans were generated with the omission of elective nodal irradiation (ENI) from the treatment portals (2D minus ENI plans [2D-ENI plans]). Both sets were inserted in the patient’s planning computed tomographies (CTs). These reconstructed plans were then compared to alternative, three-dimensional treatment plans which had been generated de novo, with the omission of ENI: 3D minus elective nodal irradiation (3D-ENI plans). Dose delivery to the planning target volumes (PTVs) and to the organs at risk were compared between the 3 sets of corresponding plans. The potential for dose escalation of each patient’s 2D-ENI and 3D-ENI plan beyond 54 Gy, standard to CHART, was also determined. Results: PTV coverage was suboptimal in the 2D CHART and the 2D-ENI plans. Only in the 3D-ENI plans did 100% of the PTV get >95% of the dose prescribed (i.e., 51.5 Gy [51.3–52.2]). Using 3D-ENI plans significantly reduced the dose received by the spinal cord, the mean and median doses to the esophagus and the heart. It did not significantly reduce the lung dose when compared to 2D-ENI plans. Escalation of the dose (minimum >1 Gy) with optimal PTV coverage was possible in 55.5% of patients using 3D-ENI, but was possible only in 16.6% when using the 2D-ENI planning strategy. Conclusions: 3DCRT is fundamental to achieving optimal PTV coverage in NSCLC. A policy of omission of elective nodal irradiation alone (and using 2D technology) will not achieve optimal PTV coverage or dose escalation. 3DCRT with omission of ENI can achieve true escalation of CHART in 55.5% of tumors, depending on their site and N-stage. © 1999 Elsevier Science Inc. Three-dimensional, conformal radiation therapy, Lung cancer, CHART, Elective nodal irradiation, Hyperfrac- tionation, Acceleration. INTRODUCTION Lung cancer is the major cause of cancer-related death in western countries (1–3). Persistence of thoracic disease and the emergence of previously occult, distant metastases con- tinually undermine therapy for patients with unresectable, locally advanced, non-small cell lung cancer (NSCLC) (3). The inadequacy of conventional radiation techniques for local treatment was clearly demonstrated by Arriagada et al. (4). Despite this, when conventional radiation techniques are combined with chemotherapy, a modest increase in survival has been demonstrated in most studies (5–10), suggesting that a combination of both modalities is the best current strategy for curative therapy for locally advanced NSCLC (11). It has been demonstrated that the therapeutic ratio can be improved with three-dimensional conformal radiation ther- apy (3DCRT), by both improved delivery to the tumor, and reduced dose to organs at risk (12,13). This has facilitated dose escalation, with resulting survival figures that may be equivalent to those achieved with combined modality treat- ment (14). In addition, advances in the understanding of tumor radiobiology have led to the development of a con- tinuous, hyperfractionated and accelerated radiation therapy scheme (CHART) for NSCLC. In a major randomised trial of CHART for NSCLC, survival was significantly improved when compared to conventional radiation technique and dose: 29% versus 20% at 2 years (15). It is possible that efficacy of radiation therapy for NSCLC Reprint requests to: Dr. C. McGibney, Medical Board Office, St. Luke’s Hospital, Highfield Road, Dublin 6, Ireland. E-mail: [email protected] This study was supported by a grant from the Irish Cancer Society, 5, Northumberland Road, Dublin 4, Ireland. Accepted for publication 16 December 1998. Int. J. Radiation Oncology Biol. Phys., Vol. 45, No. 2, pp. 339 –350, 1999 Copyright © 1999 Elsevier Science Inc. Printed in the USA. All rights reserved 0360-3016/99/$–see front matter 339

Upload: independent

Post on 29-Nov-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

PII S0360-3016(99)00095-4

CLINICAL INVESTIGATION Lung

DOSE ESCALATION OF CHART IN NON-SMALL CELL LUNG CANCER:IS THREE-DIMENSIONAL CONFORMAL RADIATION THERAPY

REALLY NECESSARY?

CAROL MCGIBNEY, F.R.C.S.I.,* OLA HOLMBERG, M.SC.,† BRENDAN MCCLEAN, PH.D.,†

CHARLES WILLIAMS , F.R.C.R.,‡ PAMELA MCCREA, M. ECON. SC, PHIL SUTTON, D.C.R.(T),§ AND

JOHN ARMSTRONG, M.R.C.P.I., D.A.B.R.*

Departments of *Radiation Oncology,†Physics,‡Radiology, and§Therapeutic Radiography, St. Luke’s Hospital, Dublin, Ireland

Purpose: To evaluate, preclinically, the potential for dose escalation of continuous, hyperfractionated, acceler-ated radiation therapy (CHART) for non small-cell lung cancer (NSCLC), we examined the strategy of omissionof elective nodal irradiation with and without the application of three-dimensional conformal radiation technol-ogy (3DCRT).Methods and Materials: 2D, conventional therapy plans were designed according to the specifications of CHARTfor 18 patients with NSCLC (Stages Ib, IIb, IIIa, and IIIb). Further plans were generated with the omission ofelective nodal irradiation (ENI) from the treatment portals (2D minus ENI plans [2D-ENI plans]). Both sets wereinserted in the patient’s planning computed tomographies (CTs). These reconstructed plans were then comparedto alternative, three-dimensional treatment plans which had been generatedde novo, with the omission of ENI:3D minus elective nodal irradiation (3D-ENI plans). Dose delivery to the planning target volumes (PTVs) and tothe organs at risk were compared between the 3 sets of corresponding plans. The potential for dose escalation ofeach patient’s 2D-ENI and 3D-ENI plan beyond 54 Gy, standard to CHART, was also determined.Results: PTV coverage was suboptimal in the 2D CHART and the 2D-ENI plans. Only in the 3D-ENI plans did100% of the PTV get>95% of the dose prescribed (i.e., 51.5 Gy [51.3–52.2]). Using 3D-ENI plans significantlyreduced the dose received by the spinal cord, the mean and median doses to the esophagus and the heart. It didnot significantly reduce the lung dose when compared to 2D-ENI plans. Escalation of the dose (minimum>1 Gy)with optimal PTV coverage was possible in 55.5% of patients using 3D-ENI, but was possible only in 16.6% whenusing the 2D-ENI planning strategy.Conclusions: 3DCRT is fundamental to achieving optimal PTV coverage in NSCLC. A policy of omission ofelective nodal irradiation alone (and using 2D technology) will not achieve optimal PTV coverage or doseescalation. 3DCRT with omission of ENI can achieve true escalation of CHART in 55.5% of tumors, dependingon their site and N-stage. © 1999 Elsevier Science Inc.

Three-dimensional, conformal radiation therapy, Lung cancer, CHART, Elective nodal irradiation, Hyperfrac-tionation, Acceleration.

INTRODUCTION

Lung cancer is the major cause of cancer-related death inwestern countries (1–3). Persistence of thoracic disease andthe emergence of previously occult, distant metastases con-tinually undermine therapy for patients with unresectable,locally advanced, non-small cell lung cancer (NSCLC) (3).The inadequacy of conventional radiation techniques forlocal treatment was clearly demonstrated by Arriagadaet al.(4). Despite this, when conventional radiation techniquesare combined with chemotherapy, a modest increase insurvival has been demonstrated in most studies (5–10),suggesting that a combination of both modalities is the bestcurrent strategy for curative therapy for locally advancedNSCLC (11).

It has been demonstrated that the therapeutic ratio can beimproved with three-dimensional conformal radiation ther-apy (3DCRT), by both improved delivery to the tumor, andreduced dose to organs at risk (12,13). This has facilitateddose escalation, with resulting survival figures that may beequivalent to those achieved with combined modality treat-ment (14). In addition, advances in the understanding oftumor radiobiology have led to the development of a con-tinuous, hyperfractionated and accelerated radiation therapyscheme (CHART) for NSCLC. In a major randomised trialof CHART for NSCLC, survival was significantly improvedwhen compared to conventional radiation technique anddose: 29% versus 20% at 2 years (15).

It is possible that efficacy of radiation therapy for NSCLC

Reprint requests to: Dr. C. McGibney, Medical Board Office, St.Luke’s Hospital, Highfield Road, Dublin 6, Ireland. E-mail:[email protected]

This study was supported by a grant from the Irish CancerSociety, 5, Northumberland Road, Dublin 4, Ireland.

Accepted for publication 16 December 1998.

Int. J. Radiation Oncology Biol. Phys., Vol. 45, No. 2, pp. 339–350, 1999Copyright © 1999 Elsevier Science Inc.Printed in the USA. All rights reserved

0360-3016/99/$–see front matter

339

can be improved by the simple combination of 3DCRTwith the radiobiologically-based, standard CHART regime,which includes elective nodal irradiation. It is also possiblethat the theoretical beneficial effect of the combination ofthese two radiation approaches may be increased furtherthrough dose escalation, by adopting an alternative policy inthe management of uninvolved lymph nodes.

The objective of this preclinical study was to compare thefeasibility of dose escalation of CHART, by using conven-tional 2D, radiation techniques combined with the omissionof elective nodal irradiation (2D-ENI), and comparing thedose escalation possible with the dose escalation achievablewhen using 3DCRT, combined with a similar policy ofomission of elective nodal irradiation (3D-ENI).

METHODS AND MATERIALS

Patient characteristicsBetween July 1996 and July 1997, the data of 18 patients

who underwent radical radiation therapy for NSCLC,Stages Ib & IIb (n 5 5) and Stages IIIa & IIIb (n 5 13)was assessed (16). All patients had radiographically visibleprimary tumor and/or nodal disease. Patients were excludedif they had stage IV disease or if their Karnofsky Perfor-mance Status (KPS) or age would have prevented theirtreatment with radical intent.

Patient details are summarized in Table 1. There were 8females and 10 males. The median age was 66 years, witha range of 54–83. All had a KPS of$70. Seventeen of the18 patients had squamous cell carcinoma, and 1 patient hadadenocarcinoma.

Each patient underwent standard, two-phase, radiationtherapy as prescribed by their radiation oncologist: (Phase I,

anterior and posterior fields; and Phase II, oblique treatmentportals). During initial simulation, the patients were immo-bilized in the supine position with their arms raised abovetheir heads, supported in vacuum polystyrene immobiliza-tion bags. The patients then underwent a planning CT scanwhich was used to plan their Phase II treatment. In Phase I,actual treatment was commenced post-initial simulation,using 2D technology. Their Phase II treatment was com-menced post verification of the proposed 3D, Phase II plan.

Study methodology: 2D CHART simulation and planningThe initial simulation film, used for the patient’s actual

treatment, was also used for designing the study treatmentportals as per CHART specifications (17). The target for the2D CHART, Phase I, included the tumor and the mediasti-num, and was based on information from the simulationfilms, as well as from diagnostic X-rays and diagnostic CTscans only. The mediastinum was defined as extending fromthe suprasternal notch to 3 cm below the carina, and in-cluded the paratracheal and ipsihilar lymph nodes, but notthe contrahilar lymph nodes. Both the mediastinum and theprimary tumor and involved nodes, with a 1-cm margin,were included in the large treatment volume, with the areaof the field at the isocenter not exceeding 240 cm2 (17), andthe magnification factor being taken into account in thedesign of the portals.

Further simulation films for the oblique, off the cord field,were taken at the time of the verification procedure of thePhase II, of each patient’s actual treatment plan. 2DCHART, Phase II, portals for use in this study were thendrawn on the oblique simulator films, with the targets basedon the simulation films’ data, as well as diagnostic CT scansdata only, without reference to the planning target volume

Table 1. General and tumor features of group (n 5 18)

Pt. no. Stage T NPTVcm3 Site Side

Age(years)

GenderM/F KPS

Weight loss(%)

1 Ib T2 N0 46 P R 62 F 80 nil2 Ib T2 N0 95 P L 70 M 70 nil3 Ib T2 N0 131 P R 62 M 80 nil4 IIIa T2 N2 266 C R 76 F 80 nil5 IIb T2 N1 210 C R 77 M 70 nil6 IIb T3 N0 94 C R 57 M 80 # 167 IIIa T3 N2 343 C R 71 M 70 nil8 IIIa T2 N2 341 C L 68 M 70 # 79 IIIb T4 N2 327 CH R 76 F 80 # 5

10 IIIb T3 N3 397 C R 71 F 90 nil11 IIIa T3 N2 291 CH L 57 F 90 nil12 IIIa T3 N2 523 A R 58 M 70 nil13 IIIa T3 N2 615 A R 54 M 90 nil14 IIIa T3 N2 466 A L 83 M 80 # 515 IIIb T4 N2 715 A L 62 M 80 # 916 IIIa T3 N1 454 A R 60 F 80 # 517 IIIa T3 N2 527 A R 64 F 80 nil18 IIIa T3 N2 445 A R 69 F 80 # 5

P 5 peripheral; C5 central; CH5 central but high; A5 apical, PTV5 planning target volume: (GTV plus involved nodes)1 1-cmmargin; T5 tumor stage;N 5 nodal stage.

340 I. J. Radiation Oncology● Biology ● Physics Volume 45, Number 2, 1999

(PTV) as drawn on the (preceding) planning CT scan. Thetarget covered included the primary tumor and involvednodes, with a 1-cm margin, the area of the field at thisseparate isocenter not exceeding 140 cm2. During simula-tion, radio-opaque catheters were placed on the center of thereference fields for the patient’s actual treatment, and facil-itated the isolation of the appropriate CT slice in the pa-tient’s planning CT scans which pertained to the center ofthe study Phase I and Phase II treatment portals.

Individual 2D CHART (Phase I and Phase II), plans weregenerated using transverse contours derived from each pa-tient’s planning CT as above. CHART specifications fordose prescription for each phase, for the spinal cord dose,and for the use of wedges were strictly adhered to. Thevariation of dose to the PTV was to remain#10%. Correc-tion factors for the dose distribution was to be made fortransmission through the lung. This was achieved by retain-ing the lung contours on the single copy plans. A dose of37.5 Gy, in 25 fractions of 1.5 Gy, was prescribed to themidplane in the large, Phase I volume, and 16.5 Gy in 11fractions of 1.5 Gy were similarly prescribed to the smallertreatment volume. Spinal cord dose was usually less than 40Gy, and never exceeded 44 Gy (17).

Once both phases of the 2D CHART plans were gener-ated, they were combined and reinserted into the corre-sponding patient’s planning CT as a reconstructed 1-phase,4-field, plan. The proposed lead shielding, initially drawnon the simulation films, was also added to these recon-structed 2D CHART plans. Proportional analysis of themonitor units used per beam in each phase in the original,2D CHART plan facilitated correct beam weighting duringreconstruction of the 2D CHART plans in the 3D system.The 54 Gy prescribed to the reconstructed plan was nor-malized to the normalization point of the larger, Phase Ivolume, which had received the greater proportion of theprescribed dose.

2D-ENI planningWhen designing the 2D-ENI targets for each patient, the

new target of primary tumor and involved nodes was out-lined on the original simulation films, anterior and oblique,with reference to diagnostic CT scans only. New treatmentportals were redrawn around the targets, allowing for anadditional margin as laid down by standard CHART proto-col (17). Therefore, both Phase I and Phase II includedtumor and involved nodal areas only, with a 1-cm margin,as visualized on the original simulation films only.

To generate the 2D-ENI, Phase I and Phase II, plans, thetransverse contours derived from each patient’s planningCT were again chosen, the CT slice chosen for the contourfor each phase being at the center of the correspondingstudy treatment fields, based on the simulation films only.Lung outlines were retained on these contours and CHARTspecifications for generation of plans and dose prescriptionfor each phase were adhered to. Once each phase of the2D-ENI plan was completed, they were combined and re-constructed in each patient’s corresponding planning CT

scan as a 1-phase, 4-field plan, with the appropriate weight-ing derived from the data on the monitor units for eachbeam in the original 2D-ENI plans, as before. Lead shields,designed on the simulation films, were again added to thereconstructed plans.

3D-ENI planningWhen patients underwent planning CT scans, CT slices

were taken from the level of the cricoid to the inferiorborder of the second lumbar vertebra. (Phillips TomoscanSR 5000 CT scanner). CT slices at 10-mm intervals weretaken throughout apart from the area of interest (for exam-ple, primary tumor and involved nodes), where the CTslices were taken at 5-mm intervals. The CT data wastransferred to the initial computer terminal of the planningsystem (Helax-TMS™ Treatment management PlanningSystem, Version 3, Helax AB, Uppsala, Sweden).

The PTV and volumes of interest, such as lungs, spine,heart and esophagus, were outlined using lung and medias-tinal window settings. The gross tumor volumes (GTVs),which included visible tumor and involved nodes only, wereoutlined initially. A 1-cm margin, which included a 1-cmcranial and caudal extension, was then added in order tomatch the specifications for the target definition in theCHART protocol (17). This margin—the integrated safetymargin—was to include the margin for the clinical targetvolume, CTV, and a margin for set-up deviations and pa-tient movement (18), thus forming the PTV. Individual,manually constructed, 0.5-cm margins for the CTV and forthe set-up margin were not undertaken in this study. Theywould now be feasible, however, using the automated mar-gin tool available in Version 4 of the Helax-TMS™ Treat-ment Management Planning System. Lymph nodes wereconsidered involved if their size was greater than 1 cm in allnodal stations except station 7 (i.e., sub-carinal, for whichthe upper limit of normal was 1.5 cm) (19,20). Those lymphnodes which were not involved were excluded from treat-ment portals, again diverging from the specifications as laiddown by the standard CHART protocol. Thus, plans createdto conform to these targets were designated as 3D minuselective nodal irradiation (3D-ENI).

The PTVs, based on gross tumor, involved nodes, andadded margins, varied from 46 cm3 to 715 cm3, with themedian of 342 cm3. Tumor sites varied within the lung, andwere designated as apical, peripheral, central or central-high, according to their position on coronal sections taken atthe isocenters of the planning target volumes. Central-hightumors were#1.5 cm from the apex of the upper lobes.

Three-dimensional radiation plans (1-phase, with a doseprescription of 54 Gy to the ICRU reference point) werecreatedde novoto conform to the PTVs in the patient’splanning CT scans. The beam’s eye view (BEV) and mul-tiplanar reconstruction facilities were used to fully encom-pass the planning target volume and to minimize dose tonormal tissue. Wedges were used in cranio-caudal and intransverse directions as necessary.

341Dose escalation of CHART in NSCLC● C. MCGIBNEY et al.

Comparison of the three planning methods and doseescalation

PTV coverage and dose to organs at risk using each of thethree planning approaches (i.e., 2D CHART, 2D-ENI, and3D-ENI) were assessed and compared as the 2D CHARTand the 2D-ENI plans had been reconstructed separately as4-field plans in the same patient files where the 3D-ENIplans had been generated subsequently in the 3D planningsystem. PTV parameters included the mean prescribed doseto PTV (percentage and Gy), minimum dose recorded in thePTV (Gy), and the percentage volume of the PTV receivingthe minimum recommended ICRU-50 dose (i.e., 95% of theprescribed dose of 54 Gy) (21).

Dose to the lungs was assessed for the total lung volume.Parameters included the mean dose received by the totallung volume (percentage and Gy), percent total lung volumereceiving 25 Gy or more (Vol. 25) and the effective lungvolume at 54 Gy. This last parameter is produced using thehistogram reduction method of Kutcheret al. to define theequivalent volume of lung receiving 54 Gy (22,23). Thisvolume was then assessed as a percentage of the totalvolume of the lung for each patient. This percentage wasused as the limiting factor for escalation. Maximum dosereceived by the spinal cord was compared for all three plans.The maximum, mean and median doses received by theesophagus and heart were also recorded.

The potential for dose escalation in each patient’s 2D-ENIand 3D-ENI plans was assessed using the reconstructed 2D-ENI plans and 3D-ENI plans. Escalation of each patient’soriginal dose (54 Gy) was continued until one of the followinglimits was exceeded: the tolerance of lung (defined in thisstudy as the effective volume level of the original 2D CHARTplan), tolerance of spinal cord as specified in the CHARTprotocol, that is, the maximum cord dose should not normallyexceed 40 Gy, and must not exceed 44 Gy (17), or a maximalpoint dose 60 Gy in the esophagus.

Statistical analysisEach patient acted as his/her own control. The SPSS

(version 8) for Windows was used to analyze the data. Thepaired t-test was utilized for comparison of corresponding

results for the parameters of PTV coverage and dose toorgans at risk. McNemar’s test for dependant paired datawas used to compare the feasibility of nominal and true doseescalation between the two planning approaches, 2D-ENIand 3D-ENI. The assessment of the potential impact ofindividual patient and tumor features on degree of doseescalation possible with 3D-ENI planning was restricted tounivariate analysis, using Kendall’s correlation test. Thenumbers within each stratum of the relevant variables weresuch that multivariate analysis was not possible.

RESULTS

Dose delivery to the PTVThe mean percentage of the prescribed dose of 54 Gy

recorded in the PTV in the 18 tumors, achieved by the 3planning approaches, varied between 97.7% and 102.3%.The mean percentage of the prescribed dose was signifi-cantly reduced by an average of 4.6% in the 2D CHARTplans when compared to the corresponding 2D-ENI plansand, also significantly reduced by an average of 3.6% whencompared to the 3D-ENI plans (p 5 0.0001,p 5 0.003,respectively) (Table 2).

The range of values for the mean percentage and mean Gyof prescribed dose also indicated that PTV coverage wassuboptimal in the 2D and 2D-ENI plans (Table 2 & Fig. 1).This was further illustrated by the minimum dose (Gy) resultsrecorded in any part of the PTVs (Table 2 and Fig. 2).

There was no significant difference in any of the param-eters of dose delivery to the PTV when 2D CHART werecompared to the corresponding 2D-ENI plans (Table 2); nordid the 2D CHART or the 2D-ENI plans achieve adequatecoverage of the PTV, according to ICRU specifications(Fig. 3). The percent volume of the PTV receiving 95% ofthe prescribed dose or more (Vol. 95) was reduced by14.48% (p 5 0.00001) in the 2DCHART plans whencompared to 3D-ENI plans. When 2D-ENI and 3D-ENIplans were compared for the same parameter, the Vol. 95was reduced in the 2D-ENI plans by 16.9% (p 5 0.0002).

Table 2. Comparison of dose to PTV

PTV

Median (range)

2D CHART 2D-ENI 3D-ENI

Percent* of prescribed dose to PTV 97.7 (85.2–102.6) 98.7 (73.4–103.4) 102.3 (101.5–104)Dose mean i.e. mean dose of PTV (Gy) 52.78 (46–55.4) 53.29 (39.6–55.7) 55.23 (54.81–56.16)Minimum dose recorded in any part of

PTV (Gy) 21.57 (2.4–47.8) 18.49 (2.4–49.7) 51.5 (51.3–52.2)Percent volume of PTV receiving

. 95% prescribed dose 87.38 (62–97) 88.5 (50–98) 100 (100–100)

PTV 5 planning target volume: (GTV1 involved nodes)1 1-cm margin; 2D-ENI5 2D technology with omission of elective nodalirradiation; 3D-ENI5 3D conformal radiation therapy with omission of elective nodal irradiation.

* Mean % of prescribed dose received by PTV of each tumor.Mean i.e. mean dose to PTV (Gy)

342 I. J. Radiation Oncology● Biology ● Physics Volume 45, Number 2, 1999

Only in the 3D-ENI plans did 100% of the PTV get$95%of the dose prescribed, median (range): 51.48 Gy (51.3–52.16), (Table 2).

Dose to organs at riskLung.As expected, the percent of the total lung volume

receiving 25 Gy or more (Vol. 25) and effective volume at54 Gy, were significantly reduced in 2D-ENI plans and inthe 3D-ENI plans when these parameters were compared tothe results in the standard 2D CHART plans, (p 5 0.01,p 5 0.001, andp 5 0.014, p 5 0.013, respectively)(Table 3). However, the mean dose to the lungs differedsignificantly only between the 2D CHART and the 2D-ENIplans (p 5 0.01).

2D-ENI plans did not differ significantly from the 3D-ENIplans in any of the three parameters of lung dose (Table 3).

EsophagusWhen 2D-ENI and 3D-ENI plans were compared, the

maximum point dose to the esophagus was significantlygreater in the 3D-ENI plans, (p 5 0.03). Themedianpercentages of the prescribed dose, received by the esoph-agus, for the 2D-ENI and the 3D-ENI plans were not

significantly different (median and range): 34.6% (2.1–70.4) and 37.2 (2.1–80.2), respectively (p 5 0.9).

HeartThe maximum dose received by the heart in the 2D-ENI

plans, was, on average, 9.6% (5.2 Gy) less when comparedto 2D CHART plans and 10.67% less when compared to the3D-ENI plans (Table 3). Neither the mean nor medianvalues for the percentage of prescribed dose to the heartwere significantly different on comparison of the 2D-ENIand the 3D-ENI plans: 11.6% (1.5–100.7) and 13.15%(1–66.8), p5 0.5 ottus and 5.25% (1.6–74.5) and 4.05%(1–63.1),p 5 0.23 ottus, respectively.

Spinal cordThere was no difference in maximum dose to the spinal

cord between standard, 2D CHART plans and the 2D-ENIplans: 39.57 (35.1–41.74) Gy and 39.71 (2.32–43.8) Gy.Maximum dose received by the spinal cord was signifi-cantly less in 3D-ENI plans when compared to either thestandard CHART 2D or the 2D-ENI plans, with meanreductions of 38.5% and 28.6%, respectively (p 5 0.00001andp 5 0.0005)(Table 3).

Fig. 1. Comparison of PTV mean doses in the three planning approaches: 2D CHART, 2D-ENI, and 3D-ENI. PTV5planning target volume: (GTV1 involved nodes)1 1-cm margin. Patients 1–3: peripheral tumors; 4–11: centraltumors; 12–18: apical tumors.

343Dose escalation of CHART in NSCLC● C. MCGIBNEY et al.

Dose escalationOn first examination, neither of the 2 approaches facili-

tated dose escalation beyond 54 Gy significantly better thanthe other (p 5 0.125) though dose escalation appearedpossible in 15/18 (83%) patients when using 2D-ENI plansbut was possible only in 10/18 (55.5%) patients using the3D-ENI approach (Table 4). There was no significant dif-ference between the degree of escalation possible overall orwhen only those patients in whom escalation was possiblewere compared when the 2 techniques were compared (Ta-ble 4).

However, when PTV coverage was assessed at the esca-lated dose levels, it was evident that the dose incrementscited as possible in the 15 escalated 2D-ENI plans werenominal and did not take suboptimal PTV coverage, whenusing this technique, into account. The suboptimal PTVcoverage was evident from the range of results for the meandose to the PTV and from the minimum dose recorded inany part of the PTV, at the escalated dose prescription(Table 4). The mean dose to the PTV in the 3D-ENI planswas satisfactory and the minimum dose percentage recordedin the PTV was 98.7%.

The nominal increments, visible in Fig. 4 were thereforeredefined as “real” increments by ensuring that 100% of the

PTV received either 54 Gy (Fig. 5) or 51.3 Gy (Fig. 6), thelatter being 95% of 54 Gy, the minimum percentage ofprescribed dose recommended by ICRU 50. Analyzed inthis way, 2D-ENI planning techniques led to a de-escalationfor all but the peripherally-sited tumors (Fig. 5). It was clearthat with 3D-ENI planning, the PTV of all the tumors,regardless of site, received 95% of the prescribed dose (i.e.,51.3 Gy). While dose escalation was not possible in allcases with 3D-ENI, particularly in those with tumors sitedin the apices of the lungs (patient nos. 12–18, Fig. 6) doseescalation was still significantly more feasible than with the2D-ENI approach, (p 5 0.0001).

Factors which influence the possibility of escalation in3D-ENI plans

In the 3D-ENI plans, there was a strong negative corre-lation between the real increments possible above 51.3 Gyand the following parameters: planning target volume cm3,T-stage and N-stage, (correlation coefficient, kt5 20.66,p 5 0.0001; kt5 20.52,p 5 0.0003; kt5 20.4,p 5 0.01,respectively). In addition, the site of the tumor also influ-enced the potential for dose escalation. Escalation was sig-nificantly greater in those with peripheral tumors than in

Fig. 2. Comparison of minimum doses recorded in any part of the PTV in the three planning approaches: 2D CHART,2D-ENI, and 3D-ENI. PTV5 planning target volume: (GTV1 involved nodes)1 1-cm margin. Patients 1–3:peripheral tumors; 4–11: central tumors; 12–18 apical tumors.

344 I. J. Radiation Oncology● Biology ● Physics Volume 45, Number 2, 1999

those with central, central-high or apical tumors, mean(SD): 58.6 Gy (5.63), 8.1 Gy (5.7), 0.4 Gy (0.46), 0.58 Gy(0.62), respectively, ANOVA,p 5 0.0001.

The impact of nodal involvement on the potential fordose escalation was heavily influenced by the inclusion ofthe three peripheral (node negative) tumors in the study.Those with node-positive disease had significantly less es-calation then those with node-negative disease—median,

range: 0.8 Gy (0–16.2) versus 52.9 Gy (3.8–69.7),p 50.004, respectively. For those with node-positive disease,the median number of lymph node stations affected was 3,range (2–7). Despite a strong positive correlation betweenthe PTV and the number of lymph node stations involved(kt 5 0.6,p 5 0.005), the negative correlation between thenumber of lymph node stations involved and the potentialfor dose escalation disappeared when those with node-

Fig. 3. Comparison of percentage of PTV receiving$95% of the prescribed dose in the three planning approaches: 2DCHART, 2D-ENI, and 3D-ENI. PTV5 planning target volume: (GTV1 involved nodes)1 1-cm margin. Patients 1–3:peripheral tumors; 4–11: central tumors; 12–18 apical tumors.

Table 3. Dose received by organs at risk (n 5 18)

Median (range)

2D CHART 2D-ENI 3D-ENI

Mean percent of 54 Gy received by total lung volume 22.9 (4.6–38.8) 19.2 (9.04–36.38) 22.3 (6.7–37.5)Mean dose to total lung volume (Gy) 12.4 (2.5–20.95) 10.4 (1.7–19.7) 12.1 (3.6–20.3)Percent volume of lung receiving$ 25 Gy 20.2 (13–36.5) 15.35 (8–34) 17.75 (4–33.52)Percent effective volume* 20.7 (12.59–36.4) 17.4 (8.85–34.08) 17.6 (5.36–33.81)Maximum point dose received by spinal cord (Gy) 39.6 (35.1–41.7) 39.7 (2.32–43.8) 19.4 (1.7–36.5)Maximum point dose to esophagus (Gy) 50.3 (5.1–56.3) 48.5 (2.2–57.2) 56.6 (2.8–59.7)Maximum point dose to heart (Gy) 54.03 (3.4–56.4) 51.5 (1.67–57.2) 55.3 (0.86–59.8)

2D-ENI 5 2D technology with omission of elective nodal irradiation; 3D-ENI5 3D conformal radiation therapy with omission ofelective nodal irradiation.

* Effective volume of lung irradiated to 54 Gy as a % of thetotal lung volume.

345Dose escalation of CHART in NSCLC● C. MCGIBNEY et al.

negative disease were excluded from the analysis: (Kt520.4,p 5 0.006 vs Kt20.2,p 5 0.13). Similarly, for thosewith node positive disease only, no significant differencecould be detected in the potential for dose escalation be-tween those with and without subcarina lymph node in-volvement. The strong positive correlation between thenumber of lymph node stations involved and the dose to theorgans at risk (lung, heart, spinal cord, and esophagus) wasalso lost when only those with node-positive disease wereanalyzed.

DISCUSSION

Rationale for 3DCRT and CHART combinationThe prognosis for patients with locally advanced, unre-

sectable NSCLC is grim, with 5 year survival figures of

,5% (1–3). Current evidence suggests that combined mo-dality therapy offers the best chance of control in those withlocally advanced disease, at high risk of occult metastases(5–10). However, the optimization of both chemotherapyand radiation modalities is still in progress (3,11,14).

Individual advances in radiation (3,13,14) have led tomodest increases in survival. In a large, prospective, mul-ticenter, randomized trial, Saunderset al. demonstrated anincrease in survival of 29% from 20% with the radiobiolo-gy-based CHART regime which addresses the repopulationof tumors by delivering a dose of 54 Gy in 12 days (15).3DCRT has been shown to improve the therapeutic ratio(12–14,25).

To determine if the efficacy of this CHART schedule fornon-small cell lung cancer (NSCLC) could be increasedfurther, we examined and compared the two following strat-

Fig. 4. Comparison of nominal increments (Nom. Inc.) above the 54 Gy of standard CHART possible in the twoplanning approaches, 2D-ENI and 3D-ENI. Patients 1–3: peripheral tumors; 4–11: central tumors; 12–18 apical tumors.

Table 4. Escalation in whole group

Median (range) Gy

2D-ENI to 2D-ENIat escalated level

n 5 15

3D-ENI to 3D-ENIat escalated level

n 5 10

Increment achieved above standard 54 Gy 4 (1–35) 11.5 (2–73)New dose (Gy) prescribed to PTV, at escalated level 58 (55–59) 65.5 (56–127)Mean dose to PTV (Gy), at escalated level 56.3 (44.8–90.7) 66.9 (57–128.9)Minimum dose to PTV (Gy) at escalated level 23.4 (2.3–81.9) 62.8 (53.3–121.03)

2D-ENI 5 2D technology with omission of elective nodal irradiation; 3D-ENI5 3D conformal radiation therapy with omission ofelective nodal irradiation; PTV5 planning target volume: (GTV1 involved nodes)1 1-cm margin.

346 I. J. Radiation Oncology● Biology ● Physics Volume 45, Number 2, 1999

egies: 2D technology with the omission of elective nodalirradiation, using standard 2D technology for planning,available in all departments (2D-ENI), and the use of3DCRT, again with the omission of elective nodal irradia-tion (3D-ENI).

Assessment parameters for these two different planningstrategies included PTV coverage and dose to organs at risk,both at baseline and at escalated levels (see Materials andMethods). A pertinent point in the methodology and assess-ment is that, from the outset, the dose to the lung sustainedin the standard 2D CHART plan was set as the limit or safeceiling dose, beyond which escalation of 2D-ENI and 3D-ENI plans could not proceed. This dose, on an individualbasis, would be as small as possible in the 2D CHARTplans, as anterior and posterior fields were used for themajor part of the 54 Gy in standard CHART regime. There-fore, the escalation of 2D-ENI, but particularly 3D-ENIplans, were limited from the outset. This was so, even if theindividual and absolute values of the lung parameters werelow, for example, in those tumors situated at the apices ofthe lungs, suggesting a low risk of pneumonitis.

The results for target coverage and for dose to organs atrisk were analyzed separately, but then integrated whensuboptimal PTV coverage in the 2D CHART and 2D-ENIplans was indicated (Table 2).

Target coverageThe mean dose to the PTV on prescription of 54 Gy to the

ICRU reference point, suggested that a policy of electivenodal irradiation omission with planning being undertaken

using 2D-ENI would have similar PTV coverage as thatachievable with 3D-ENI. This, together with the fact thatdose to the lungs was also the same in both 2D-ENI and3D-ENI plans (vide infra), would imply that 3DCRT is notnecessary to facilitate dose escalation of the CHART pre-scription. However, it was clear from the minimum dose tothe PTV and the percentage of PTV receiving$95% of thedose (Vol. 95), that adequate PTV coverage was achievedonly in the 3D-ENI plans (Table 2, and Figures 1–3).

These results indicate that a simple policy of omission ofelective nodal irradiation, even in the context of multimo-dality therapy, would be insufficient for local control ofNSCLC at baseline level of prescription. This suboptimalbaseline target coverage would further undermine any(nominal) increments for dose escalation initially thoughtpossible with this policy (Table 4). When the nominalincrements possible were adjusted to take account of theminimal dose to the PTV, the increments became signifi-cant decrements from the standard 54 Gy instead (Figures 5and 6).

Dose to organs at riskThe initial results of dose to the lungs and maximum

doses to the heart and to the esophagus suggested that3DCRT was not necessary for dose escalation of CHART:there was no difference in dose to the lungs between the2D-ENI plans and 3D-ENI plans, and the maximum dosesreceived by the heart and the esophagus were greatest in the3D-ENI plans, and least in the 2D-ENI plans.

The 2D-ENI and 3D-ENI plans did not differ in any of

Fig. 5. Comparison of real increments (Real. Increm.) possible, above/below 54 Gy, in the two planning approaches,2D-ENI and 3D-ENI. Patients 1–3: peripheral tumors; 4–11: central tumors; 12–18: apical tumors.

347Dose escalation of CHART in NSCLC● C. MCGIBNEY et al.

the lung parameters, suggesting that the addition of 3DCRTto the omission of ENI policy did not reduce the risk forpneumonitis. The effective volume of lung at 54 Gy as apercentage of the total lung volume was reduced by anaverage of 2.9% and 3.4% as one progressed from 2DCHART to either 2D-ENI or the 3D-ENI plans, respec-tively. The volume of lung receiving$25 Gy was similarlyreduced by#5% (Table 3). However, if the PTV wasadequately covered in the 2D-ENI plans, it is likely thatmore normal lung tissue would have been traversed therebycreating a greater disparity between 2D-ENI and 3D-ENIplans and resulting in an improved therapeutic ratio for lungby the combination of 3DCRT and ENI omission policywhen compared to the implementation of the omission ofENI policy alone (2D-ENI). A limitation of this study is thatthe impact of the hyperfractionation and acceleration of theregime was not addressed. The assessment of lung doseassociated with the three planning approaches was limitedto the mean dose, the Vol. 25, and the effective volume (forexample, no prediction of risk of pneumonitis using normaltissue complication probabilities could be made).

The limit of 60 Gy to the esophagus was adhered to inthis study as a result of the reported incidence of dysphagiain the trial by Saunderset al. (15). While maximum dose tothe esophagus was greatest in the 3D-ENI plans, and least inthe 2D-ENI plans (on average the dose to the esophagus inthe 3D-ENI plans was 12.5% and 16.75% greater than in 2DCHART and 2D-ENI plans, respectively), 3 points are im-portant: firstly, as in the case of the dose to the lungs, if thetreatment portals had been increased in the 2D-ENI plans to

produce adequate PTV coverage, the difference in maxi-mum dose to the esophagus might be less between 2D-ENIand the 3D-ENI plans; secondly, the volume of esophagusreceiving the maximum dose is very small and not signifi-cantly different between any of the planning strategies: themedian and range of volumes being 0.25 cm3 (0.08–3.9),0.24 cm3 (0.10–1.82), and 0.31 cm3 (0.12–8.6) for 2DCHART, 2D-ENI, and 3D-ENI plans, respectively. The 8.6cm3 upper limit in the range for the 3D-ENI plans occurredin the individual dose bin of 1.62 Gy in 1 of the peripheraltumors. With an increase in length in the treatment portalsof the 2D-ENI plans to accommodate suboptimal PTVcoverage, even the small disparity in volume of esophagusirradiated at the upper dose levels between the 2D-ENI andthe 3D-ENI plans would have been less. Thirdly, both themean and the median values for the dose received by theesophagus were significantly smaller than the maximumdose recorded, the smallest median values being recorded inthe 3D-ENI plans (see Results section on esophagus).

On assessment of maximal point dose (and the mean andthe median doses), received by the spinal cord, as expectedand unlike the other organs at risk, the therapeutic ratio wassignificantly improved using 3D-ENI plans when comparedto standard CHART and the 2D-ENI plans (Table 3). Suchan increase in the therapeutic ratio would be significant forthe reduction in symptoms of transient radiation myelopathycharacterized by L’hermitte’s sign (24), which occurred in42 of 338 patients at follow-up in the CHART arm of therandomized trial of Saunder’set al. (15).

Fig. 6. Comparison of real increments (Real. Increm.) possible in the two planning approaches 2D-ENI and 3D-ENIabove/below 51.3 Gy. Patients 1–3: peripheral tumors; 4–11: central tumors; 12–18 apical tumors.

348 I. J. Radiation Oncology● Biology ● Physics Volume 45, Number 2, 1999

Dose escalationThis was possible for all the peripherally-sited tumors

(patients 1–3) in both 2D-ENI and 3D-ENI plans. For thesetumors, both the nominal and real increments, assessed atboth 54 Gy or 51.3 Gy, were greater in the 3D-ENI plans(Figs. 5 and 6).

When using the elimination of ENI policy alone (i.e., inthe 2D-ENI plans), no real escalation was possible for eithercentrally placed (patients 4–11) or apical tumors (patients12–18) when the suboptimal PTV coverage was accountedfor (Fig. 5). With ENI elimination and 3DCRT technology,mean escalation ($1 Gy) above the recommended dose(95% of prescribed dose) in 10 of the 18 patients facilitatedan increase to a new dose, median (range): 66.9 Gy (56–128.9).

Feasibility of escalation was influenced by T-stage, N-stage and by the size of the PTV. While multivariate anal-ysis could not be undertaken to determine the individual andindependent contribution of these factors some features ofthe influence of the N-stage were clear. The impact ofN-stage on escalation was due to the difference in escalationbetween those with node-negative vs node-positive disease(p 5 0.004) only. Neither the number of lymph node sta-tions involved nor the presence or absence of subcarinalinvolvement influenced the feasibility of escalation, oncethe 4 N0tumorswere excluded from the analysis.

Tumor site also impacted on the feasibility of escalation:of those in whom escalation was possible, only 1 was anapically-placed tumor (Fig. 6). This is in contrast to thereport of Grahamet al. who have demonstrated clinicallythat pneumonitis risk varies with site of tumor and is actu-ally reduced in the upper lobes when compared to the lowerlobes (25). This highlights a further limitation of this study,for example, the use of the maximum lung dose of theCHART plan as the ceiling dose beyond which escalation

cannot proceed. In the standard 2D CHART plans, themaximum percentage of the dose prescription is given AP/PA. Dose to the lung is lowest in those with apical tumors,because of this and because the normal tissue irradiated inthese tumors is restricted to the inferolateral border. Thisresults in the ceiling for lung dose being reached quiterapidly in these tumors when escalating the corresponding3D-ENI plans. However, the results also demonstrated thatfor apically-sited tumors, the absolute lung doses are lowerthan that recorded in tumors located at other sites. Thissuggests that dose escalation may be still possible if otherlimits for dose escalation are chosen, for example, escala-tion to a point where the absolute lung dose in the apicaltumors reaches the median level of those with central tu-mors (provided this level was not associated with pneumo-nitis).

CONCLUSION

A policy of elimination of elective nodal irradiation (andusing standard 2D technology) is insufficient for dose esca-lation, being undermined by the associated suboptimal tar-get coverage.

The prospects of enhanced local control and improvedsurvival with the CHART treatment regimen may be in-creased with 3DCRT with the omission of ENI through theability of improved therapeutic ratio and facilitation of doseescalation beyond 54 Gy.

The relative merits of two strategies for dose escalationhave been tested in this study using a lung ceiling basedupon the 2D CHART plans for each patient. It may bepossible to escalate beyond the total doses reached in thisstudy by defining an alternative ceiling for lung dose; this isparticularly so for apically sited tumors.

REFERENCES

1. Ginsberg R, Kris M, Armstrong JG. Non-small lung cancer. InDeVita VT, Hellman S, and Rosenberg SA, editors. Cancer:Principles and practice of oncology. 4th ed. Philadelphia:Lippincott; 1993.

2. Schottenfeld D. Epidemiology of lung cancer. In: Pass HI,Mitchell JB, Johnson DH,et al., editors. Lung cancer: prin-ciples and practice. Philadelphia: Lippincott-Raven Co.; 1995.p. 305–323.

3. Perez CA, Pajak TF, Rubin P,et al. Long-term observation ofthe patterns of failure in patients with unresectable non-oatcell carcinoma of the lung treated with definitive radiotherapy:Report by the Radiation Oncology Group.Cancer1987;59:1874–1881.

4. Arriagada R, Le Chevalier T, Quoix E,et al. Astroplenary:Effect of chemotherapy on locally advanced non-small celllung carcinoma: A randomized study of 353 patients.Int JRadiat Oncol Biol Phys1991;20:1183–1190.

5. Burkes R, Ginsberg R, Shepherd F,et al. Neo-adjuvant trialwith MVP (Mitomcin-c 1 vindesine1 cisplatin) chemother-apy for stage III (T1–3, N2, M0) unresectable non-small celllung cancer (NSCLC).Proc Am Soc Clin Oncol1989;8:221.

6. Dillman RO, Seagren SL, Propert KJ,et al. Randomised trial

of induction chemotherapy plus high dose radiation versusradiation alone in stage III non-small-cell lung cancer.N EnglJ Med1990;323:940–945.

7. Sause WT, Scott C, Taylor S,et al. Radiation Therapy On-cology Group (RTOG) 88–08 and Eastern Cooperative On-cology Group (ECOG) 4588: Preliminary results of a phase IIItrial in regionally advanced, unresectable non-small-cell lungcancer.J Natl Cancer Inst1995;87:198–205.

8. Schaake-Konig C, Van den Bogaert W, Dalesio O,et al.Effects of concomitant cisplatin and radiotherapy on inoper-able non-small-cell lung cancer.N Engl J Med1992;326:524–530.

9. Le Chevalier T, Arriagada R, Quoix E,et al. Radiotherapyalone versus combined chemotherapy and radiotherapy innonresectable non-small-cell lung cancer: First analysis of arandomized trial in 353 patients.J Natl Cancer Inst1991;83:417–423.

10. Non-small Cell Lung Cancer Collaborative Group. Chemo-therapy in non-small cell lung cancer: A meta-analysis usingupdated data on individual patients from 52 randomized clin-ical trials.Br Med J1995;311:899–909.

11. Kirkbride P, Gelmon K, Eisenhauer E,et al. A phase I/II study

349Dose escalation of CHART in NSCLC● C. MCGIBNEY et al.

of Paclitaxel (Taxol) and concurrent radiotherapy in advancednon-small-cell lung cancer.Int J Radiat Oncol Biol Phys1997;39:1107–1111.

12. Armstrong JG. Three-dimensional conformal radiotherapy:Precision treatment of lung cancer.Chest Surg Clin North Am1994;4:29–43.

13. Armstrong JG, Zelefsky MJ, Leibel SA,et al. Strategy fordose escalation using 3-dimensional conformal radiation ther-apy for lung cancer.Ann Oncol1995;6:693–697.

14. Armstrong JG, Raben A, Zelefsky M,et al. Promising survivalwith three-dimensional conformal radiation therapy for non-small cell lung cancer.Radiother Oncol1997;44:17–22.

15. Saunders M, Dische S, Barrett A,et al. Continuous hyperfrac-tionated accelerated radiotherapy (CHART) versus conven-tional radiotherapy of non-small cell lung cancer: A random-ised multicentre trial.Lancet1997;350:161–165.

16. International Union Against Cancer. Lung and pleural tumors.In: Sobin LH, Wittekind CH, editors. TNM classification ofmalignant tumors. 5th Edition New York: Wiley-Liss, Inc.;1997. p. 91–100.

17. CHART Steering Committee: A trial comparing conventionalfractionation with CHART in the radical treatment of non-small cell carcinoma of the bronchus. January 1990.

18. Ekberg L, Holmberg O, Wittgren L,et al. What marginsshould be added to the clinical target volume in radiotherapy

treatment planning for lung cancer?Radiother Oncol1998;48:71–77.

19. Glazer HS, Aronberg DJ, Sagel SS,et al. CT demonstration ofcalcified mediastinal lymph nodes: A guide to the new ATSclassification.AJR1986;147:17–20.

20. Murray JG, Breatnach E. The American Thoracic Societylymph node map: A CT demonstration.Eur J Radiol1993;17:61–68.

21. International Commission on Radiation Units and Measure-ments: Report 50. Prescribing, recording, and reporting pho-ton beam therapy. Bethesda, MD: September, 1993.

22. Kutcher J, Burman C. Calculation of probability factors fornon-uniform normal tissue irradiation. The effective volumemethod.Med Phys1987;14:487.

23. Kutcher GJ, Burman C, Brewster L,et al. Histogram reductionmethod for calculating complication probabilities for three-dimensional treatment planning evaluations.Int J Radiat On-col Biol Phys1991;21:137–146.

24. Jones A. Transient radiation myelopathy (with reference toL’hermitte’s sign of electrical parasthesia).Br J Radiol1964;37:727–744.

25. Graham MV, Purdy JA, Emami B,et al. Preliminary resultsof a prospective trial using three dimensional radiotherapyfor lung cancer.Int J Radiat Oncol Biol Phys1995;33:993–1000.

350 I. J. Radiation Oncology● Biology ● Physics Volume 45, Number 2, 1999