Int. J. Radiation Oncology Biol. Phys., Vol. 80, No. 2, pp. 608–613, 2011Copyright � 2011 Elsevier Inc.

Printed in the USA. All rights reserved0360-3016/$ - see front matter

robp.2010.07.1994

doi:10.1016/j.ij

PHYSICS CONTRIBUTION

A METHOD FOR THE PREDICTION OF LATE ORGAN-AT-RISK TOXICITYAFTERRADIOTHERAPY OF THE PROSTATE USING EQUIVALENT UNIFORM DOSE

CATHY FLEMING, M.SC.,* COLIN KELLY, M.SC.,y PIERRE THIRION, M.D.,z

KATHRYN FITZPATRICK, B.SC.,* AND JOHN ARMSTRONG, M.D.z

From the *Clinical Trials Unit and Departments of yPhysics and zRadiation Oncology, St. Luke’s Hospital, Rathgar, Dublin Ireland

ReprinPhysics -BoulevardCFlemingThis w

Purpose: To evaluate the predictive value of equivalent uniform doses (EUD) for late bladder and rectal toxicityafter high-dose three-dimensional conformal radiation therapy (3D-CRT) to the prostate.Materials and Methods: Using the method developed by Kutcher et al., EUDs for whole bladder and rectum werecalculated from the dose�volume histograms of 180 patients with localized prostate cancer treated to 70�74 Gywith 3D-CRT. Late complications were recorded using the Radiation Therapy Oncology Group scale, correlatedagainst EUD and known physical predictive indicators.Results: EUD is an independent prognostic factor for Grade 2+ long-term rectal and bladder toxicity after radi-ation treatment to the prostate. Patients receiving an EUD >63.1 Gy to the rectum have a statistically significant(10% vs. 30%; p = 0.002) higher risk of developing Grade 2+ late complications. Patients receiving an EUD >53.4Gy to the bladder have a statistically significant (10% vs. 33%; p = 0.001) higher risk of developing Grade 2+ latecomplications.Conclusions: It has been demonstrated that EUD is a strong independent predictive factor for Grade 2+ late com-plications after 3D-CRT to the prostate. Threshold values have been demonstrated for both bladder and rectum,above which there is a clinically significant increased risk of complications. � 2011 Elsevier Inc.

Late rectal toxicity, Late bladder toxicity, Prostate, Equivalent uniform dose (EUD), 3D-CRT.

INTRODUCTION

First three-dimensional conformal radiotherapy (3D-CRT)and now intensity-modulated radiotherapy (IMRT) have al-lowed the delivery of doses above 78 Gy to become com-monplace in the radical treatment of prostate cancer (CaP).Such dose escalation has been shown to result in a significantincrease in local control (1–5). The limiting factor inachieving further dose escalation has been identified as thedose to the organs at risk (OARs). In the case of CaP,these dose-limiting organs are represented mainly by rectumand bladder; however, small bowel, femoral heads, and pe-nile bulb may also be considered (6).

Much work has been conducted into developing tools forthe prediction of both early (7) and late (8–10) toxicities.The goal of such work is to allow for dose escalation whilereducing the incidence of toxicity. Studies in this area haveconcentrated on the analysis of dose�volume histogram(DVH) data obtained from treatment planning systems.Significant correlations have been reported between laterectal and bladder toxicities for a wide range of DVHparameters (i.e., V50, V60, V70, and V75 [% of organ

t requests: Cathy Fleming, M.Sc., Department of ImagingUnit 1472, MDAnderson Cancer Center, 1515 Holcombe, Houston, TX 77030-4409. Tel: 713-792-3841; E-mail:1@mdanderson.orgork was made possible by grants awarded from the Irish

608

receiving 50–75 Gy]) (8, 10–17). An alternative approachhas been suggested using the concept of equivalent uniformdose (EUD) (10, 18–20), a mathematical concept thatassumes two dose distributions are equivalent if they causethe same radiobiological effect (21). To date, EUD hasbeen used as a tool to compare treatment plans (22, 23) ormathematically as an optimization parameter in inverseplanning systems (19, 24�26). In this article, we evaluatethe use of EUD as an independent prognostic factor forlong-term rectal and bladder toxicity after 3D-CRT for CaP.

METHODS AND MATERIALS

Trial protocol and eligibilityThe data from 275 patients included on the Ireland Cooperative

Oncology Research Group 97/01 (27) trial were recorded. Patientsreceived either 4 or 8 months of hormonal treatment together with3D-CRT. Seventy patients were later excluded for breaches of eli-gibility criteria, 40 of these elected to be treated in a private facility,16 were excluded for undergoing hip replacements, 5 subsequentlyrefused RT, 2 died before or during RT, and the remaining 7 had nofollow-up visit. Twenty-five patients were removed from this studyfor technical reasons related to archiving of plans or charts.

Cancer Society (ICS) and the St Luke’s Institute for Cancer Re-search (SLICR).Conflict of interest: none.Received May 4, 2010, and in revised form June 26, 2010.

Accepted for publication July 16, 2010.

Eud and Late Toxicity d C. FLEMING et al. 609

All patients received conformal localized radiation therapy to theprostate and seminal vesicles. Dose prescription was 70 Gy in 35fractions (58 patients), 73.68 Gy in 35 fractions (118 patients), or69.5 Gy in 33 fractions (4 patients), in which the dose is prescribedto the isocenter (the 73.68 Gy and 69.5 Gy schemes correspond to70 Gy and 66 Gy, respectively, prescribed to the 95% isodose, aswas practiced by one physician). The 4 patients receiving the lowerprescription did so because they did not meet the dose�volumeconstraints. The planning target volume was obtained by addingan isocentric margin of 1 cm around the CTV, except posteriorlyin the area of the anterior rectal wall where the margin was reducedto 0.8 cm. No patient received pelvic node irradiation. The majority(166/180) of patients were treated in a supine position. Dose pre-scription to the planning target volume followed the recommenda-tions of International Commission on Radiation Units andMeasurements 50 (28) in regard to coverage and homogeneity.At the time our institutional class solution for prostate irradiationconsisted of a three field-box technique, using an anterior andtwo lateral fields.To minimize prostate and seminal vesicle positional variability,

institutional protocol specified that patients undergo the planningcomputed tomography with a full bladder (6 cups of water andwait for 30 min) and an empty rectum, treatment was under thesame conditions. Both the bladder and rectum were defined asthe whole organ, including contents. The median (� SD) bladdervolume was 282.3 � 198.9 mL, whereas the respective values forthe rectum were 90.8 � 53.0 mL. Constraints were applied suchthat no more than 50% of the rectal volume should receive morethan 50 Gy and the maximum dose to the rectum should neverexceed 75 Gy; there were no constraints applied to the bladder inthis trial.

Equivalent uniform doseIndividual DVHs for bladder and rectum were obtained from the

treatment planning system (Helax TMS v6.0, MDS Nordion,Ottawa, Canada) for all patients. The DVHs are then reduced toa single index to allow direct comparison (21, 29�31) accordingto the method outlined by Niemierko (21).

Each individual dose element (Di) and its associated volume el-ement (DVi) are taken from the DVH and translated, using the his-togram reduction method of Kutcher (20, 30), into a single volumeelement irradiated to a reference dose. This method assumes thateach volume element of the differential DVH independentlyobeys the same dose�volume relationship as the whole organ(20). An effective volume (Veff) at a reference dose (Dm) is obtainedby scaling each individual volume element (Vi) appropriately ac-cording to its associated dose element (Di), then summing the resul-tant through the following equation:

Veff ¼X�

Di

Dm

�1n

DVi

This single bin of the differential DVH corresponds to a uniform cu-mulative DVH with dose equal to a reference dose (Dm). The EUDfor whole organ irradiation is that dose (Dm) at which the ratio ofVeff to Vref is unity. This is calculated by means of a simple C++program, designed and verified in-house, which takes as an inputthe DVHs in the format exported by the treatment planning systemwithout any modification. The values of n were taken as 0.12 (rec-tum) and 0.5 (bladder) as given by Burman et al. (32) and takenfrom tolerance data compiled by Enami et al. (13).

End points and statistical analysisFor the 180 patients, median follow-up was 59 months (inter-

quartile range 45�71 months). End points were Grade 2+ long-term bladder and rectal toxicity. Toxicity was scored using the Ra-diation Therapy Oncology Group and the European Organizationfor Research and Treatment of Cancer scale (33). The late toxicitygrade was defined as the maximum grade occurring and the cumu-lative complication probability is used. A recursive partitioningprogram (STREE [34]) was used to select the optimum point atwhich to separate the patient population into low- and high-riskgroups (using the log�rank statistic). Based on a review of the lit-erature (8–16, 35), a range of physical parameters (from V55Gy toV74Gy) along with EUD were entered into STREE (34) to selectwhich of these factors has the most predictive power. The compli-cation probabilities within each of these groups were investigatedand compared. Kaplan-Meier analysis of time to failure was under-taken using SPSS (version 15.0) with the time taken from the lastdate of treatment up to the most recent visit.

RESULTS

Physical predictive indicatorsSeveral clinical factors were first considered to ensure that

they were not confounding factors for toxicity in this patientpopulation. Neither the initial bladder nor rectal volumewere associated univariately, using independent samplest-test, with an increased risk of toxicity (bladder: p = 0.802;rectum: p = 0.484). Similarly, whether a patient had 4 vs.8 months of hormone therapy was not related to their riskof toxicity (Pearson chi-squared; bladder: p = 0.136; rectump = 0.903).

Parameters from V55 to V74 as well as the median dosewere obtained from the DVHs for each patient. As describedin End points and statistical analysis section, several ac-cepted physical predictive indicators were first investigatedto verify our patient population. Recursive partitioning anal-ysis was then used to assess cutoff values, abovewhich prob-ability of a Grade 2+ complication occurring is greater.There were significant differences in the complication ratesbetween the low- and high-risk groups for each of theseparameters in the bladder and most of the parameters inthe rectum as shown in Table 1.

Equivalent uniform doseThe median EUD to the rectum was 62.6 Gy (range, 54.1–

70.7 Gy), the median EUD to the bladder was 44.4 Gy(range, 18–65.4 Gy). Recursive partitioning analysis se-lected 63.1 Gy for the rectum and 53.4 Gy for the bladderas the optimum points at which to divide the populationsinto low- and high-risk groups. Figures 1 and 2 shows theKaplan-Meier curves for the cumulative risk of long-termGrade 2+ rectal and bladder complications over time, withthe population split at these points. They show a significantdifference in the complication rates between these twogroups (rectum; 10% vs. 30%; p = 0.002: bladder; 10% vs.33%; p = 0.001).

The predictive value of the various physical parametershas been demonstrated through univariate analysis as de-scribed in the Physical predictive indicators section. They,

Fig. 2. Kaplan-Meier curve showing the cumulative probability ofGrade 2+ bladder complications over time: divided at an equivalentuniform dose of 53.38 Gy.

Table 1. Table showing the cutoff points and p values for allparameters investigated

Bladder Rectum

Parameter Cutoffp Value

(log–rank) Cutoffp Value

(log–rank)

V55 78.1% 0.002* 2.1% 0.013*V60 65.1% 0.005* 37.8% 0.055V65 48.5% 0.007* 38.3% 0.004*V70 33.4% 0.002* 36.9% 0.002*V72 32% 0.001* 35.5% 0.006*V74 30.8% 0.002* 34.4% 0.006*Median dose 44.4% 0.013* 30.3% 0.111Equivalent uniform dose 53.4% 0.000* 63.1% 0.002*

* Indicates statistical significance at the p < 0.05 level using thelog–rank test.

610 I. J. Radiation Oncology d Biology d Physics Volume 80, Number 2, 2011

along with EUD, were entered into a multivariate analysis toselect which of these parameters is the most powerful at pre-dicting Grade 2+ late complications after radiation therapy.In the case of both the bladder and the rectum, the most pow-erful predictor (i.e., the first split in the tree) was the EUD.

DISCUSSION

Organ delineation and fillingTo minimize inter-fraction variability of bladder volume,

patients were instructed to undergo computed tomographywith a full bladder (6 cups of water and wait for 30 min),which was also requested during treatment. Pinkawa et al.(36) have demonstrated that the ability of patients to fill theirbladder remains unchanged at the end of radiation treatment.Patients were instructed to empty their rectum before theplanning computed tomography and before each treatment.Stasi et al. (37) have demonstrated that this simple practicereduces the impact of organ motion on the dose�volumeparameters of the rectum. Since this study has finished, theuse of rectal balloons in an effort to ensure reproducibility

Fig. 1. Kaplan-Meier curve showing the cumulative probability ofGrade 2+ rectal complications over time: divided at an equivalentuniform dose of 63.14 Gy.

of both rectal size and position has become more common(38–40).

Physical dose constraints and treatment planningMany studies have correlated a wide variety of points on

cumulative DVHs with the incidence of late radiation se-quelae and they are used throughout the radiation oncologycommunity as the criteria for acceptance or rejection oftreatment plans. A review of literature on dose�volume ef-fects in the pelvis has recently been published by Fiorinoet al. (17). The first action of our study was to investigatethese points and verify that our population was representa-tive of that seen in other studies. The predictive values ofa range of these physical parameters (V55Gy–V74Gy) are val-idated this study.

During the plan evaluation process, the method by whicha physician selects a treatment plan may be quite subjective,depending on both the experiences of the physician, institu-tional practice and published evidence-based estimates ofthe cure and complication rates. Because of the wide numberof verified physical predictive indicators, there is much de-bate within medical circles as to which of these factors areto be considered and also with respect to the relative impor-tance of these factors. There are a number of issues raised byusing these single physical points as dose constraints. First,if there are two plans, both of which satisfy some, but not all,of the physical criteria, how do we select which plan is ‘‘bet-ter’’? Certain criteria are more important depending on howserial or parallel the organ is, but which is more important forthe prostate V70 or V75? Second, how can you differentiatebetween two different treatment plans, both of which satisfyall the physical criteria at the given points but that have rad-ically different DVHs? An empirical method of differentiat-ing the ‘‘better’’ plan between these two is necessary. This iswhat led us to consider using a method, such as EUD, thatincorporates information from the entire DVH as an addi-tional tool in the treatment planning process.

Eud and Late Toxicity d C. FLEMING et al. 611

Correlation of EUD with toxicityPrevious work on this patient population, presented at ES-

TROBarcelona 2007 (41), focused on treating EUD as a con-tinuous variable vs. toxicity. A linear quadratic model wasfitted to the complication data for years one and three afterradiotherapy. However, these models are not very practica-ble for daily use in the clinic where most institutionaldose�volume criteria employ threshold values. The estab-lishment of a simple value for daily clinician use was themotivation behind the work in this article.

Recursive partitioning analysis selected 63.1 Gy for therectum and 53.4 Gy for the bladder as the optimum pointsat which to divide the populations into low- and high-riskgroups. Above these threshold EUD values, the complicationprobabilities are significantly higher (rectum; p = 0.002:bladder; p = 0.001). As well as being statistically significant,the differences in complication probabilities above and be-low these thresholds (rectum; 10% vs. 30%: bladder; 10%vs. 33%) are at a level that is significant within a clinicalsetting. To validate the use of EUD, we had to compare itspredictive power with that of the previously verified param-eters. This was done by means of recursive partitioning anal-ysis using STREE (34). For both the bladder and the rectum,the first split in the recursive tree (i.e., the most powerfulpredictor, using the log–rank statistic) was the EUD.

The value of n for the rectum of 0.12 is given by Burmanet al. (37). There is much discussion regarding the values ofthese volume parameters. A study by Dale et al. (42)showed that Normal Tissue Complication Probability(NTCP) became more significant as n was reduced, corre-sponding to a more serial rectum. On the other hand, Ran-cati et al. showed that an n value of 0.23 (68% CI: -0.09,+0.19) was a best fit to their population for Grade 2+ bleed-ing (43), suggesting that the rectum may have a more paral-lel architecture. When they considered more severebleeding (Grade 3+), the best fit for n was found to be0.06, suggesting that the rectum may behave as a more se-rial organ for the end point of severe bleeding. A review ofthe literature by Michalski et al. (16) established a value forn of 0.09 with the 95% confidence interval ranging from0.04 to 0.14. What is clear is that more investigation mustbe done into these parameters, particularly if they are tobe used in any model which attempts to ‘‘truly’’ modelthe radiobiology rather than in a phenomenological modelsuch as EUD.

EUD has so far been used to compare treatment plans(22, 23), or as an optimization parameter in treatmentplanning software (19, 24–26), rather than being directlycorrelated with late toxicity. MacKay et al. (44) haveused biologically effective dose and normal dose�surfacehistograms to predict late rectal complication after hypo-fractionated prostate radiotherapy (50 Gy in 16 fractions).An EUD of #74 Gy to the rectum was proposed as a limitby Belderbos et al. however this was derived from theconstraint of Veff <30%, which has been used in thedose escalation trial (45) rather than from experimentaldata.

Inverse treatment planningIn IMRT, dose�volume constraints are not just criteria for

the acceptance of a treatment plan or to allow dose escala-tion. With inverse planning, we optimize around these phys-ical points, possibly leading to different dose distributionsthan those seen with 3D-CRT. The standard physical doseconstraints that we use are, in general, derived from studieswhere patients were treated 3D-CRTand, although they maystill be valid with the changing dose distributions seen withIMRT, other points may be more relevant. In addition, ina situation where inverse radiation treatment planning is be-ing used the optimization process is satisfied after the vari-ous preset physical dose�volume constraints have beenmet and no further attempt is made by the optimization en-gine to improve on the plan. Qi et al. (26) have compareda number of different treatment planning systems that use bi-ologically based optimization over a number of differentsites (brain, head and neck, lung, pancreas, and prostate).They demonstrated that biological optimization resulted inplans that delivered improved sparing of normal tissuewhen compared with standard dose�volume optimizationwhere both planning techniques used the same beam ar-rangement. Das (25) took a slightly different approach bystarting with conventional dose–volume objectives, fixingcertain key points on the DVH and using EUD in an attemptto further improve on the plan. In his prostate example, theadditional biological optimization reduced the EUD to thebladder and rectum by 5% and 3.9%, respectively. Rijkhorstet al. (46) have used EUD to evaluate the effect of online mo-tion corrections on both CTV coverage and rectal dose. Theydemonstrated that the use of online correction resulted ina rectal EUD (n = 0.13) that was in all cases smaller thanplanned.

Using EUD in clinical practiceCalculation of EUD as described in this work is a simple

process. The DVH exported by the treatment planning sys-tem requires nomanipulation of the file format or content be-fore being read in by a simple in-house C++ program.Calculations of EUD and biologically based optimizationtechniques have become commonly available on commer-cial treatment planning software in the last number of years.QUANTEC (46), a comprehensive review of the issues sur-rounding the use of models in the clinic, discusses many ofthe issues surrounding the practical application of modelssuch as EUD. One of the major benefits of the concept ofEUD is that its calculation is independent of prescriptionor method of delivery; it provides a simple method for com-paring plans from different treatment modalities. In doingthis, an area of concern is the difference in dose distributionsproduced between 3D-CRT, IMRT, and proton treatment.Trofimov et al. (23) simultaneously planned 10 prostate pa-tients with both IMRTand 3D conformal proton therapy andthey used EUD to evaluate dose to the bladder and rectum.Using an n value for the rectum of 0.2 (CI 0.33–0.125) and0.143 (CI 0.25–0.083), they found no significant differencein the EUDs to the bladder and rectum for the two treatment

612 I. J. Radiation Oncology d Biology d Physics Volume 80, Number 2, 2011

techniques. Techniques such as IMRT and 3D conformalproton therapy also provide for dose-escalation above thatwhich has been commonplace until recently. Delivery ofdoses in the region of 81 Gy by means of IMRT has becomecommonplace. Even though these techniques are more con-formal, small areas of the rectum adjacent to (or within) theplanning target volume may receive higher doses than be-fore. In using dose–volume parameters, both physical andmodel-based, we need to be cognizant of the evolving natureof dose distributions to normal tissues as treatment tech-niques advance.

CONCLUSION

This study has shown that EUD is a strong predictiveindicator of late rectal and bladder toxicity after radiation

therapy to the prostate. The threshold values identified inthis study, an EUD of 63.1 Gy for the rectum and 53.4 Gyfor the bladder, divide the population into risk groups thatare not only statistically significant but that give differencesin complication probability that are significant in the clinicalsetting.

It is important to note that we are not advocating thediscontinuation of the use of physical constraints for planevaluation. Physical dose constraints have long been investi-gated and their use validated. We are suggesting that theEUD concept outlined in this article be used as a complementto these existing physical predictive indicators. In particular,EUD could prove effective in cases where it is not possible tosimultaneously satisfy all physical dose constraints or tocompare two plans, both of which are acceptable accordingto institutional protocol.

REFERENCES

1. Hanks GE, Hanlon AL, Epstein B, et al. Dose response in pros-tate cancer with 8–12 years’ follow-up. Int J Radiat Oncol BiolPhys 2002;54:427–435.

2. Lyons JA, Kupelian PA, Mohan DS, et al. Importance of highradiation doses (72 Gy or greater) in the treatment of StageT1-T3 adenocarcinoma of the prostate. Urology 2000;55:85–90.

3. Pollack A, Hanlon AL, Horwitz EM, et al. Radiation therapydose escalation for prostate cancer: A rationale for IMRT.World J Urol 2003;21:200–208.

4. Pollack A, Smith LG, von Eschenbach AC. External beam ra-diotherapy dose response characteristics of 1127 men withprostate cancer treated in the PSA era. Int J Radiat OncolBiol Phys 2000;48:507–512.

5. ShipleyWU, Verhey LJ, Munzenrider JE, et al. Advanced pros-tate cancer: The results of a randomized comparative trial ofhigh dose irradiation boosting with conformal protons com-pared with conventional dose irradiation using photons alone.Int J Radiat Oncol Biol Phys 1995;32:3–12.

6. Fisch BM, Pickett B, Weinberg V, et al. Dose of radiation re-ceived by the bulb of the penis correlates with risk of impotenceafter three-dimensional conformal radiotherapy for prostatecancer. Urology 2001;57:955–959.

7. Pollack A, Zagars GK, Starkschall G, et al. Conventional vs.conformal radiotherapy for prostate cancer: Preliminary resultsof dosimetry and acute toxicity. Int J Radiat Oncol Biol Phys1996;34:555–564.

8. Fiorino C, Sanguineti G, Cozzarinin C, et al. Rectal dose-volume constraints in high-dose radiotherapy of localizedprostate cancer. Int J Radiat Oncol Biol Phys 2003;5:953–962.

9. Schulthesis TE, LeeWR, HuntMA, et al. Late GI and GU com-plications in the treatment of prostate cancer. Int J Radiat OncolBiol Phys 1997;37:3–11.

10. Wachter S, Gerstner N, Goldner G, et al. Rectal sequelae afterconformal radiotherapy of prostate cancer: Dose-volumehistograms as predictive factors. Radiother Oncol 2001;59:65–70.

11. Boersma LJ, Van Den Brink M, Bruce AM, et al. Estimation ofthe incidence of late bladder and rectum complications afterhigh-dose (70–78 Gy) conformal radiotherapy for prostate can-cer, using dose-volume histograms. Int J Radiat Oncol BiolPhys 1998;41:83–92.

12. Cozzarini C, Fiorino C, Luca Ceresoli G, et al. Significant cor-relation between rectal DVH and late bleeding in patients

treated after radical prostatectomy with conformal or conven-tional radiotherapy (66.6–70.2 Gy). Int J Radiat Oncol BiolPhys 2003;55:688–694.

13. Emami B, Lyman J, Brown A, et al. Tolerance of normal tissueto therapeutic irradiation. Int J Radiat Oncol Biol Phys 1991;21:109–122.

14. Hung EH, Pollack A, Levy L, et al. Late rectal toxicity: Dose-Volume effects of conformal radiotherapy for prostate cancer.Int J Radiat Oncol Biol Phys 2002;54:1314–1321.

15. Kupelian PA, Reddy CA, Carlson TP, et al. Dose/volume rela-tionship of late rectal bleeding after external beam radiotherapyfor localized prostate cancer: Absolute or relative rectal vol-ume? Cancer J 2002;8:62–66.

16. Michalski JM, Gay H, Jackson A, et al. Radiation dose-volumeeffects in radiation-induced rectal injury. Int J Radiat OncolBiol Phys 2010;76(3 Suppl):S123–S129.

17. Fiornio C, Valdagni R, Rancati T, et al. Dose-volume effects fornormal tissues in external radiotherapy: Pelvis. Radiother On-col 2009;93:153–167.

18. Ebert MA. Viability of the EUD and TCP concepts as reliabledose indicators. Phys Med Biol 2000;45:441–457.

19. Wu Q,Mohan R, Niemierko A, et al. Optimization of intensity-modulated radiotherapy plans based on equivalent uniformdose. Int J Radiat Oncol Biol Phys 2002;52:224–235.

20. Kutcher GJ, Burman C, Brewster MS. Histogram reductionmethod for calculating complication probabilities for three-dimensional treatment planning evaluations. Int J Radiat OncolBiol Phys 1991;21:137–146.

21. Niemierko A. Reporting and analyzing dose distributions: Aconcept of equivalent uniform dose. Med Phys 1997;24:103–107.

22. Jones LC, Hoban PW. Treatment plan comparison using equiv-alent uniform biologically effective dose (EUBED). Phys MedBiol 2000;45:159–170.

23. Trofimov A, Hguyen PL, Coen JJ, et al. Radiotherapy treat-ment of early-stage prostate cancer with IMRT and protons:A treatment planning comparison. Int J Radiat Oncol BiolPhys 2007;69:444–453.

24. Schwartz M, Lebesque JL, Mijnheer BJ, et al. Sensitivity oftreatment plan optimisation for prostate cancer using the equiv-alent uniform dose (EUD) with respect to the rectal wall vol-ume parameter. Radiother Oncol 2004;73:209–218.

25. Das S. A role for biological optimization within the currenttreatment planning paradigm. Med Phys 2009;36:4672–4682.

Eud and Late Toxicity d C. FLEMING et al. 613

26. Qi XS, Semenenko VA, Li XA. Improved critical structuresparing with biologically based IMRT optimization. MedPhys 2009;36:1790–1799.

27. Armstrong J, Taylor J, Thirion P, et al. Preliminary results ofa randomised trial comparing short vs. protracted NeoadjuvantHormonal Therapy (NHT) prior to radiation therapy (RT) of lo-calised prostate cancer. (Irish Clinical Oncology ResearchGroup Protocol # 97/01). Radiother Oncol 2004;73(S1):S78.

28. ICRU Report 50, Prescribing, Recording and Reporting PhotonBeam Therapy. Bethesda, MD; 2000.

29. Hamilton CS, Chan LY, McElwain DLS, et al. A practical eval-uation of five dose-volume histogram reduction algorithms.Radiother Oncol 1992;24:251–260.

30. KutcherGJ, BurmanC. Calculation of complication probabilityfactors for non-uniform tissue irradiation: The effective volumemethod. Int J Radiat Oncol Biol Phys 1989;16:1623–1630.

31. Lyman JT, Wolbarst AB. Optimisation of radiation therapy, III:A method of assessing complication probabilities from dose-volume histograms. Int J Radiat Oncol Biol Phys 1987;13:103–109.

32. Burman C, Kutcher GJ, Emami B, et al. Fitting of normal tissuetolerance data to an analytic function. Int J Radiat Oncol BiolPhys 1991;21:123–135.

33. Cox JD, Stentz J, Pajak TF. Toxicity criteria of the RadiationTherapy Oncology Group (RTOG) and the European Organiza-tion for Research and Treatment of Cancer (EORTC). Int J Ra-diat Oncol Biol Phys 1995;31:1341–1346.

34. Zhang HP, Singer B. Analysis of censored data: survival trees.In: Zhang HP and Singers. Recursive Partitioning in HealthSciences. New York: Springer, 1999;93–101.

35. Fiorino C, Cozzarini C, Vavassori V, et al. Relationships be-tween DVHs and late rectal bleeding after radiotherapy forprostate cancer: Analysis of a large group of patients pooledfrom three institutions. Radiother Oncol 2002;64:1–12.

36. Pinkawa M, Asadpour B, Gagel B, et al. Prostate position var-iability and dose-volume histograms in radiotherapy for pros-tate cancer with full and empty bladder. Int J Radiat OncolBiol Phys 2006;64:856–861.

37. Stasi M, Munoz F, Fiorino C, et al. Emptying the rectum beforetreatment delivery limits the variations of rectal dose-volume

parameters during 3DCRTof prostate cancer. Radiother Oncol2006;80:363–370.

38. Bastasch MD, The BS, Mai WY, et al. Tolerance of endorectalballoon in 396 patients treated with intensity-modulated radia-tion therapy (IMRT) for prostate cancer. Am J Clin Oncol 2006;29:8–11.

39. van Lin EN, Kristinsson J, Philipens ME, et al. Reduced laterectal mucosal changes after prostate three-dimensional con-formal radiotherapy with endorectal balloon as observer inrepeated endoscopy. Int J Radiat Oncol Biol Phys 2007;67:799–811.

40. The BS, Woo SY, Mai WY, et al. Clinical experience with in-tensity modulated radiation therapy (IMRT) for prostate cancerwith the use of rectal balloon for prostate immobilization.MedDosim 2002;27:105–113.

41. Fleming C, Kelly C, Thirion P, et al. A Method usingequivalent uniform dose for the prediction of late organ-at-risk toxicity following radiotherapy of the prostate(ICORG 97–01). Radiother Oncol 2007;84(Suppl 1) [Poster,abstract 344].

42. Dale E, Olsen DR, FossA SD. Normal tissue complicationprobabilities correlated with late effects in the rectum afterprostate conformal radiotherapy. Int J Radiat Oncol BiolPhys 1999;43:385–391.

43. Rancati T, Fiorino C, Gagliardi G, et al. Fitting late rectalbleeding data using different NTCP models: Results from anItalian multi-centric study (AIROPROS0101). Radiother On-col 2004;73:21–32.

44. MacKay RI, Hendry JH, Moore CJ, et al. Predicting late rectalcomplications following prostate conformal radiotherapy usingbiologically effective doses and normalised dose-surface histo-grams. Br J Radiol 1997;70:517–526.

45. Belderbos J, DeJaeger K, Lebesque JV, et al. First results ofa Phase I/II dose escalation trial in non-small cell lung cancerusing three-dimensional conformal radiotherapy. RadiotherOncol 2003;66:119–126.

46. Bentzen SM, Constine LS, Deasy JO, et al. Quantitative Anal-yses of Normal Tissue Effects in the Clinic (QUANEC): An in-troduction to the scientific issues. Int J Radiat Oncol Biol Phys2010;76(3 Suppl):S3–S9.