Consolidating duodenal and small bowel toxicity data via isoeffective dose calculations based on compiled clinical data

Practical Radiation Oncology (2014) 4, e125–e131

www.practicalradonc.org

Original Report

Consolidating duodenal and small bowel toxicity data via isoeffective dose calculations based on compiled clinical data Phillip Prior PhD, An Tai PhD, Beth Erickson MD, X. Allen Li PhD ⁎ Department of Radiation Oncology, Medical College of Wisconsin, Milwaukee, Wisconsin Received 9 November 2012; revised 22 March 2013; accepted 6 May 2013

Abstract Purpose: To consolidate duodenum and small bowel toxicity data from clinical studies with different dose fractionation schedules using the modified linear quadratic (MLQ) model. A methodology of adjusting the dose–volume (D,v) parameters to different levels of normal tissue complication probability (NTCP) was presented. Methods and Materials: A set of NTCP model parameters for duodenum toxicity were estimated by the χ 2 fitting method using literature-based tolerance dose and generalized equivalent uniform dose (gEUD) data. These model parameters were then used to convert (D,v) data into the isoeffective dose in 2 Gy per fraction, (DMLQED2,v) and convert these parameters to an isoeffective dose at another NTCP (DMLQED2’,v). Results: The literature search yielded 5 reports useful in making estimates of duodenum and small bowel toxicity. The NTCP model parameters were found to be TD50(1) model = 60.9 ± 7.9 Gy, m = 0.21 ± 0.05, and δ = 0.09 ± 0.03 Gy - 1. Isoeffective dose calculations and toxicity rates associated with hypofractionated radiation therapy reports were found to be consistent with clinical data having different fractionation schedules. Values of (DMLQED2’,v) between different NTCP levels remain consistent over a range of 5%-20%. Conclusions: MLQ-based isoeffective calculations of dose–response data corresponding to grade ≥ 2 duodenum toxicity were found to be consistent with one another within the calculation uncertainty. The (DMLQED2,v) data could be used to determine duodenum and small bowel dose– volume constraints for new dose escalation strategies. © 2014 American Society for Radiation Oncology. Published by Elsevier Inc. All rights reserved.

Introduction Unresectable pancreatic cancer patients pose a significant challenge given their high risk for loco-regional and distant progression. Historically, radiation doses of 30-60 Conflict of interest: None. ⁎ Corresponding author. Department of Radiation Oncology, Medical College of Wisconsin, 8701 Watertown Plank Rd, Milwaukee, WI 53226. E-mail address: [email protected] (X. Allen Li).

Gy in 1.8-3.0 Gy per fraction (Gy/fx) have been combined with concurrent 5-fluorouracil or gemcitabine chemotherapy with limited impact, though higher doses are likely necessary to control this disease. 1-3 Gaining local control with radiation therapy (RT) is further complicated given the location and tumor bulk in the upper abdomen amid many critical organs (duodenum especially). A previous report on pancreatic dose escalation to 70-72 Gy using 3dimensional conformal RT (3DCRT) reported severe small bowel toxicity. 4 Recent dosimetry studies of

1879-8500/$ – see front matter © 2014 American Society for Radiation Oncology. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.prro.2013.05.003

e126

P. Prior et al

pancreatic dose escalation have demonstrated improved small bowel sparing using intensity modulated RT (IMRT), image guided RT, and adaptive RT techniques compared with 3DCRT. 5,6 Clinically, the linear quadratic (LQ) model has been useful in comparing the biologic effectiveness between different RT treatment schedules. The increased use of large hypofractionated RT (HFRT) associated with stereotactic body RT has renewed debate concerning the validity of the LQ model for fractionated doses larger than ~ 15 Gy. 7 Several authors have proposed modified radiobiologic models or modifications to the LQ model that account for the constant slope in cell survival curves at high doses; however, all these models’ parameter values are obtained through mathematical fitting of cell survival curves. Ideally, the model parameters would be found using data obtained from clinical measurements of isoeffect at different fractionation schedules. The major goal of this paper is to investigate the feasibility of estimating modified radiobiologic model parameters from the best available conventionally fractionated (ie, 2 Gy/fx) RT (CFRT) and HFRT dose–response data. Our current work incorporates a modified LQ (developed in reference 8) model in the Lyman-Kutcher-Burman (LKB) normal tissue complication probability (NTCP) model, where the data from clinical reports are used in mathematical fitting and testing our dose–response model of duodenum and small bowel toxicity. Additionally, we present a methodology of converting dose–response data to another NTCP level.

Methods and materials Literature search A comprehensive literature search was performed to identify studies reporting tolerance dose data, generalized equivalent uniform dose (gEUD) data, or dose–volume parameters associated with any level of duodenum and small bowel toxicity [(D,v)-toxicity]. Additional studies were collected containing μ1/2, α/β, and CyberKnife (Accuray, Hauppauge, NY) treatment delivery times applicable to our calculations. Some studies were eliminated as the superior-inferior extent of the report’s delineated structure referred to as the “small bowel” do not include the duodenum or an association between (D,v), and toxicity were not reported. When necessary, data were digitized from figures in reports using Engage Digitizer, v4.1 (digitizer.sourceforge.net).

Practical Radiation Oncology: March-April 2014

model was utilized to convert dose–response data from CFRT and HFRT reports into the same biologically effective dose 1 NTCPðt Þ ¼ pffiffiffiffiffiffi ∫t−∞ e−x2=2 dx; 2π

ð1Þ

where t is given by    t ¼ D MLQED2 −TD50 ð1Þmodel ⋅v −n = m−TD50 ð1Þmodel ⋅v −n ; ð2Þ where TD50(1) model is the dose corresponding to 50% complication probability for a uniformly irradiated whole organ, m characterizes the steepness of the NTCP curve, v is the volume fraction of the organ uniformly irradiated, n represents the volume effect, and DMLQED2 is the MLQ based isoeffective dose delivered in 2 Gy/fx producing the same radiobiologic effect as a total dose, D, delivered in d Gy/fx:   dG ðλT þ δd Þ D MLQED2 ¼ D 1 þ = α=β   2⋅G ðλT CFRT þ 2⋅δÞ 1þ ; α=β

ð3Þ

where α and β are LQ parameters, λ is the repair rate (= ln2/μ1/2, where μ1/2 is the repair half time), T is the delivery time (TCFRT and THFRT denote the times for CFRT and HFRT reports, respectively), δ was the parameter introduced to replicate constant slope in the survival curve at high doses, and G(x) = (2/x) ⋅ [1−(1/x) ⋅ (1−exp(− x))] is the dose protraction factor (where x is the argument of G(x), and is equal to either λ T + δd or λ T + 2δ). Values for μ1/2, α/β, and THFRT will be determined during the literature search, while a value of 5 min will be used for TCFRT throughout our study. The NTCP model has 4 adjustable parameters, TD50, m, n, and δ that were determined using the least χ 2 method by minimizing  2 NTC P model ðt Þ−NTC P literature ðt Þ i i N χ2 ¼ ∑i¼1 ; ð4Þ σ2i where NTCPi model (t) and NTCPi literature (t) are the NTCP for the ith data point, while σi2 is the statistical error in the ith data point. Goodness of fit is determined by χ 2/d.o.f., where d.o.f. is the degrees of freedom (total number of literature-derived data points minus the number of adjustable model parameters) and a good fit has χ 2/d.o.f.≈1.0. Optimization was performed and standard errors in parameter estimates were obtained using Microsoft Excel add-in, SolverStat (Microsoft Corp, Redman, WA). 9

Lyman-Kutcher-Burman and modified linear quadratic model

Dose–volume parameter comparison and adjustment

The LKB NTCP model was used to describe the dose– response of the duodenum and small bowel, where the MLQ

Those reports with (D,v)-toxicity data were converted to (DMLQED2,v)-toxicity data using Eq 3 and δ found by

Practical Radiation Oncology: March-April 2014

Duodenal and small bowel toxicity data

model fitting. The model will be tested by grouping the (DMLQED2,v)-toxicity data by toxicity grade and checking for overlap in calculation uncertainties. In certain circumstances, it would be advantageous to adjust DMLQED2 at a fixed volume, v, to a different NTCP; an equation can be derived by rewriting Eq 2 as D MLQED2 ¼ ð1 þ m⋅t Þ⋅T D50 ð1Þmodel: v −n

ð5Þ

Equation 5 assumes v is uniformly irradiated to Dresulting in the NTCP calculated by Eq 1. NTCP is uniquely determined by t (the number of standard deviations away from a standard normally distributed mean corresponding to an NTCP in Eq 1, and known as the standard-score/Z-score). The dose, DMLQED2′, delivered to the same volume required to achieve another NTCP level, t', can be calculated using the ratio of Eq 5 for 2 dose– volume pairs with the same volume:   1 þ m⋅t ′ ⋅DMLQED2 : D MLQED2 ′ ¼ ð6Þ ð1 þ m⋅t Þ MLQED2,

These dose values were adjusted for NTCP levels of 5%, 7%, 10%, 12%, 15%, 17%, and 20%, corresponding to the values for t of -1.645, -1.476, -1.282, -1.175, -1.036, -0.954, and -0.842. The standard deviation in DMLQED2′ and DMLQED2 are found by taking the quadrature sum of the individual fractional errors in m and δ: σDMLQED2 ′

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi    2 ∂DMLQED2 ′ ∂DMLQED2 ′ 2 2 ; ¼ σm þ σδ ∂m ∂δ ð7Þ

  ∂DMLQED2 σDMLQED2 ′ ¼ σδ ; ∂δ

e127

Burman et al 10 (2 Gy/fx) and gEUD was available in Murphy et al 11 (25 Gy/fx), which were both used for model fitting. Murphy et al, 11 Kelly et al, 12 Kopek et al, 14 and Verma et al 13 were 4 reports that provided (D,v)toxicity data. The model will be tested by comparing Murphy et al (DMLQED2’, v) to the other 3 reports. The clinical endpoint among accepted reports was a mixture of grade ≥ 2 Common Terminology Criteria (versions 3 and 4) duodenum toxicity, and the Radiation Therapy Oncology Group criteria for late morbidity. Duodenum toxicity in Verma et al consisted of duodenal ulcer, perforation, obstruction, and fistula. The endpoint for Murphy et al consisted of duodenum ulcer and gastrointestinal (GI) hemorrhage, perforation, stricture, gastroduodenal ulcer, and duodenojejunal ulcer. Duodenum toxicity in Kopek et al consisted of duodenum ulceration and stenosis. The endpoint for Burman et al was small bowel obstruction, and perforation and fistula. Duodenum toxicity in Kelly et al consisted of duodenitis, nonbleeding ulcers, duodenal bleeding without perforation, ulceration or bleeding. The endpoints for Kelly et al, Verma et al, Murphy et al, and Kopek et al were endoscopically verified. Our calculations were carried out using values chosen to be consistent with the best available clinical data in the literature. The value for α/β = 4.0 Gy is consistent with the following clinical reports: (1) Deore et al 15 reported an α/β = 3.9 Gy for bowel stricture and perforation; and (2) a value of α/β = 4.3 Gy for various bowel late effects reported by Dische et al. 16 Additionally, the value of μ1/2 = 1 hr is consistent with the 1.5-2.5 hr estimate for various pelvic complications. 17 A total treatment time of 5 min was used for Kelly et al and Verma et al, while a time of 1 and 0.5 hour was used for Murphy et al and Kopek et al, respectively, which are on the order of the 1-3 hours CyberKnife HFRT treatment times reported by Schellenberg et al. 18

ð8Þ

where σDMLQED2′, σm and σδ are model fitting standard errors in DMLQED2′, m, n, and δ, respectively, while ∂DMLQED2′/∂m and ∂DMLQED2′/∂δ are partial derivatives of DMLQED2′ with respect to variables m and δ, respectively, and ∂DMLQED2/∂δ is the partial derivative of DMLQED2 with respect to δ.

Results Literature search A comprehensive literature search of duodenum and small bowel dose–response data identified 5 reports (Table 1, refs. 10-14). The reported fractionation scheme was 1.9 Gy/fx in 1 report, 12 2.0 Gy/fx in 2 reports, 10,13 15 Gy/fx in another, 14 and 25 Gy/fx in 1 report. 15 Uniform whole organ tolerance dose data were available in

Lyman-Kutcher-Burman and modified linear quadratic model The χ 2 analysis of uniform whole organ tolerance dose (including TD50(1), TD5(1), and gEUD in Table 1) data yielded the following LKB model parameters: TD50(1) model = 60.9 ± 7.9 Gy, m = 0.21 ± 0.05, δ = 0.09 ± 0.03 Gy - 1 , and χ 2 /d.o.f. = 0.6 (P = .615). Though the tolerance dose data available prevented us from obtaining a value of n (ie, v = 1 in Eq 2), the value of n = 0.12 (95% confidence interval [CI], 0.06-0.26) derived from patient data in Murphy et al agree with the n = 0.15 from Burman et al. The model fit to the clinical data is shown in Fig 1. The error bars associated with the data were obtained from a digitized figure within Murphy et al, while those from Burman et al were estimated to be 1.7% and 25% for TD5(1) and TD50(1), respectively (TD5(1) is the dose corresponding to 5% complication probability for a uniformly irradiated whole organ).

e128 Table 1 Study

P. Prior et al

Practical Radiation Oncology: March-April 2014

Clinical data from reports meeting search criteria Primary Cancer

Burman — et al 11

Treatment Technique

No. of Dose Patients (Gy)

Dose/fx Dosimetric/Model Toxicity (Gy) Parameters (%)

Endpoints

—

—

28 a

Small intestine obstruction and perforation/fistula

Murphy Locally advanced et al 11 and surgically unresectable pancreatic cancer

73 73% CyberKnife & 27% Linacbased SBRT

Locally advanced Kelly et al 12 unresectable pancreatic cancer

73.6% 3DCRT 26.4% IMRT

Verma Para-aortic nodal et al 13 metastases

IMRT

Kopek Unresectable Linac-based et al 14 cholangiocarcinoma SBRT

—

TD5(1) = 40 Gy

5.0 ± 1.7

TD50(1) = 55 Gy m = 0.16 n = 0.15 V20 b 3.3 cc V20 ≥ 3.3 cc D1cc b 23 Gy D1cc ≥ 23 Gy V25 b 0.21 cc V25 ≥ 0.21 cc EUD: 10.5 Gy EUD: 16.6 Gy EUD: 19.6 Gy EUD: 22.6 Gy TD50(1) = 24.6 Gy m = 0.23 n = 0.12 V55 ≤ 1.0 cc V55 N 1.0 cc

50.0 ± 25.0

11.0 48.0 12.0 48.0 12.0 52.0 0.0 ± 0.0 11.0 ± 7.6 17.0 ± 8.9 37.0 ± 11.4 50%

CTC3 grade ≥ 2 duodenal ulceration, duodenal perforation, duodenal stricture, gastroduodenal ulcer, duodenojejunal ulcer, and GI hemorrhage.

9.0 47.0

CTC4 Grade ≥2 duodenal toxicity (including, duodenitis, nonbleeding ulcers, duodenal bleeding without perforation, ulceration, or bleeding) Grade ≥2 RTOG duodenum ulceration, obstruction, stenosis, perforation, and fistula Grade = 1 CTC3 Duodenal ulceration & stenosis Grade ≥ 2CTC3 Duodenal ulceration & stenosis

25

25

106

58.5 b

1.90 b

112

45-66 c 1.8-2.2 d Dmax = 63.3 Gy D2cc = 61.2 Gy D5cc = 59.1 Gy

7.0 7.0 7.0

27

45

D1cc = 25.3 Gy

55.2

D1cc = 37.4 Gy

44.8

15

CTC3, National Cancer Institute Common Terminology Criteria for Adverse Events version 3.0; CTC4, National Cancer Institute. Common Terminology Criteria for Adverse Events version 4.0; EUD, equivalent uniform dose; GI, gastrointestinal; IMRT, intensity. modulated radiation therapy; RTOG, Radiation Therapy Oncology Group late radiation morbidity scoring criteria; SBRT, stereotactic body radiation therapy; TD5(1), 5% probability of experiencing complication; TD50(1), 50% probability of experiencing complication. a A dose per fraction of 2 Gy was used for Emami et al 1991. b The average prescription dose and dose per fraction is reported (based on 50.4 Gy at 1.8Gy/fx in 78 patients and an average of 66.5 Gy in 1.99 Gy/fx in 28 patients). c Prescription dose for patients who underwent resection was 45-50 Gy, while those having gross residual disease received 45-50 Gy to the nodal CTV and a boost of 60-66 Gy. d Fractionation schedule and events scored as duodenal toxicity were disclosed by personal communication with Drs A.H. Klopp, MD (in October 2011), and J. Verma, MD (in May 2012).

The NTCP error bars in the model fitting were calculated using the following 3 references within Emami et al 19: (1) Potish et al 20 reported 7 out of 92 patients experiencing small bowel obstruction; (2) Gallagher et al 21 reported 5 out of 150 patients experiencing small bowel obstruction; and (3) Lilliemoe et al 22 reported 14 out of 17 patients experiencing small bowel obstruction. A standard error in TD5(1) was based on the number of patients

experiencing toxicity and total number of patients from Potish et al and Gallagher et al, and was calculated to be 1.7%; while the standard error in TD50(1) was based on the number of patients experiencing toxicity and total number of patients from Lilliemoe et al and was calculated to be 25.0%. Optimization using unequal weighting was used because the scatter in the data is not the same in all parts of the dose–response curve. Consequently, our model over-

Practical Radiation Oncology: March-April 2014

Duodenal and small bowel toxicity data

e129

point dose data from Verma et al associated with a 7% incidence of toxicity.

Adjustment of dose–volume-toxicity parameters

Figure 1 Lyman-Kutcher-Burman NTCP model curve. The curve represents the estimated NTCP model as function of dose. The DMLQED2 is the isoeffective dose in 2 Gy fractions. Error bars for Burman et al 10 data points (squares) represent estimated standard errors, while the error bars for Murphy et al 11 data points (diamonds) were taken from Fig 4 of their paper.

estimates the error in the tolerance data due largely to the Burman et al data as indicated by a χ 2/d.o.f. smaller than 1.0. However, a second model fitting using a nonweighted χ 2 method yielded model parameters and asymptotic confidence intervals that overlapped with those presented. Therefore, our model parameters are not overly sensitive to the error estimates and reflect the uncertainty in the best available clinical data.

Dose–volume toxicity parameters The converted dose–response data used α/β = 4.0 Gy, λ = 0.693 hr - 1, and δ = 0.09 Gy - 1 in Eq 3 to calculate (DMLQED2, v)-toxicity data for all previously mentioned reports and are listed in Table 2. The isoeffective doses ranged from 48.3-77.2 Gy and were clustered around 5966 Gy. The standard error in Table 2 represents the uncertainty in DMLQED2 due to the uncertainty in δ calculated using Eq 8. The toxicity associated with CFRT and HFRT dose− response data in Table 2 show some consistency. A 44.8% incidence of grade ≥ 2 duodenum toxicity in Kopek et al was associated with D1 cc = 77.2 ± 6.4 Gy (D1 cc is the dose delivered to 1 cc or less of duodenum and small bowel volume) and is consistent with the 48% incidence of grade ≥ 2 toxicity associated with D1cc ≥59.0 ± 5 Gy from Murphy et al. Kelly et al reported an actuarial rate of 47% that was associated with V55 N 1 cc (V55 is the absolute duodenum and small bowel volume [cm 3 (cc)] irradiated to a dose greater than or equal to 55 Gy), which is also consistent with Murphy et al’s 48% toxicity associated with D1 cc ≥ 59.0 ± 5 Gy. Last, the uncertainty in the dose–volume data associated with a 12% incidence of toxicity of V66.3 in Murphy et al overlap with the max

The data from Murphy et al 11 (Table 3A) (NTCP = 12%) and from Verma et al 13 (Table 3B) (NTCP = 7%) were selected to adjust their values of DMLQED2 to different levels of NTCP. Dose–volume parameters of duodenum and small bowel for a different NTCP level are tabulated from Table 3, where the calculation uncertainty was calculated using Eq 7. Values of DMLQED2′ for Murphy et al varied from 10.8 to 14.8 Gy over an NTCP of 5%-20%. A slightly larger variation of 14.5-15.5 Gy can be observed in the data from Verma et al. The standard error in DMLQED2′ was larger for data of Murphy et al (3.97.2 Gy) than data of Verma et al (1.0-4.2 Gy). Most notably, the Verma et al maximum point dose falls within in the calculation uncertainty Murphy et al’s 0.21 cc volume over all NTCP values. Data in Table 3 may be used to determine the dose–volume constraints corresponding to an acceptable NTCP value.

Conclusion We have proposed a method of estimating radiobiologic model parameters using clinical isoeffect data at CFRT and HFRT schedules. Using the limited number of clinical data for the duodenum, we found overlapping values of DMLQED2 within the calculation uncertainties for a series of published dose–response data (Table 2). This suggests our analysis yielded a plausible set of LKB MLQ based model parameters of duodenum toxicity. Though the value of n is inferred from the literature, it could be determined using patient-specific dose–response data from different fractionation schedules having the same endpoint. Differences in toxicity rates among converted dose− response data may be attributed to differences in scoring criteria, treatment techniques (CyberKnife has improved high dose normal tissue sparring compared with IMRT 23), method of reporting normal tissue toxicity, contouring normal tissue structures, and patient heterogeneity. Although the grading scales used by reports in Tables 1 and 2 have similar criteria, the duodenum specific toxicities were often reported in combination with other GI toxicities (such as vomiting, nausea, and anorexia). Additionally, the toxicity rates of Kelly et al 12 and Murphy et al 11 were 12month Kaplan-Meier estimates of duodenum toxicity and the other reports may underestimate the true incidence as they report crude estimates of toxicity. 24 The findings presented in our work are not without limitation. The parameter δ was chosen as a fitting parameter because clinically realistic values were found in the literature for the other parameters. Though other phenomenological radiobiologic models describe the

e130

P. Prior et al

Table 2

Practical Radiation Oncology: March-April 2014

Results of DMLQED2 calculations using modified linear quadratic model

Study

(DMLQED2,V)

Endpoints CTC3 grade ≥ 2 duodenum toxicity (11%) CTC3 grade ≥ 2 duodenum toxicity (52%) CTC3 grade ≥ 2 duodenum toxicity (12%) CTC3 grade ≥ 2 duodenum toxicity (48%) CTC3 grade ≥ 2 duodenum toxicity (12%) CTC3 grade ≥ 2 duodenum toxicity (45%) Grade ≥ 2 CTC4 duodenum toxicity (9%)

Verma et al 13

V48.3 ± 3.9 Gy b 3.3 cc V48.3 ± 3.9 Gy ≥ 3.3 cc D1 cc b 59.0 ± 5.5 Gy D1 cc ≥ 59.0 ± 5.5 Gy V66.3 ± 6.7 Gy b 0.21 cc V66.3 ± 6.7 Gy ≥ 0.21 cc V53.7 Gy ≤ 1 cc V53.7 Gy N 1 cc V58.6 Gy ≤ 0.01 cc V58.6 Gy N 0.01 cc D2cc = 61.2 Gy

Verma et al 13

D5cc = 59.1 Gy

Verma et al 13

Dmax = 63.3 Gy

Kopek et al 14 Kopek et al 14

D1 D1

Murphy et al

11

Murphy et al 11 Murphy et al 11 Kelly et al 12 Kelly et al 12

= 43.4 ± 4.1 Gy a a cc = 77.2 ± 6.4 Gy cc

Grade ≥ 2 CTC4 duodenum toxicity (47%) Grade ≥ 2 RTOG duodenum ulceration, obstruction, stenosis, perforation, and fistula (7%) Grade ≥ 2 RTOG duodenum ulceration, obstruction, stenosis, perforation, and fistula (7%) Grade ≥ 2 RTOG duodenum ulceration, obstruction, stenosis, perforation, and fistula (7%) Grade b 2 CTC3 duodenal ulceration & stenosis (55.2%) Grade ≥ 2 CTC3 duodenal ulceration & stenosis (44.8%)

Note: For Murphy et al11 duodenum toxicity includes the following: CTC3 grade ≥ 2 duodenal ulceration, duodenal perforation, duodenal stricture, gastroduodenal ulcer, duodenojejunal ulcer, and gastrointestinal hemorrhage. CTC3, National Cancer Institute Common Terminology Criteria for Adverse Events version 3.0; CTC4, National Cancer Institute Common Terminology Criteria for Adverse Events version 4.0; HFRT, hypofractionated radiation therapy; MLQED2, Modified Linear Quadratic based Equivalent Dose in 2 Gy/fx; RTOG, Radiation Therapy Oncology Group criteria for late toxicity. a DMLQED2 calculations for Kopek et al used THFRT = 0.5 hours.

effects of cell killing at large hypofractionated doses, the MLQ is advantageous given the 1 additional parameter modeling this effect. There is some debate that the radiobiologic response to RT delivered in 1 to 5 fractions could be different than the response to CFRT. 7 However, in the context of abdominal stereotactic body RT, Murphy et al 11 reported histologic data suggesting that large single fraction RT and CFRT cause abdominal tissue damage through similar mechanisms. The tolerance doses available in Burman et al 10 are the best available for small bowel toxicity although they are largely unchallenged. Nonetheless, our calculations sugTable 3 Volume at dose

gest that they may reasonably describe duodenum and small bowel dose–response given the good agreement between converted dose–response data in Table 2. Lastly, the (DMLQED2',v) data in Table 3 strictly apply to uniform whole organ irradiation or gEUD data and some caution should be exercised as these numerical estimates should not be substituted for clinical judgment. Table 3 could be used as planning constraints during IMRT optimization while inhomogeneous dose distributions could be handled using the effective volume method proposed by Ten Haken et al 25 (using the TD50(1) and m reported here and n = 0.12) to escalate the dose.

Dose–volume parameters at various levels of normal tissue complication probability (NTCP) NTPC = 5%

NTCP = 7%

NTCP = 10%

NTCP = 12%

NTCP = 15%

NTCP = 17%

NTCP = 20%

DMLQED2′ (Gy) 41.9 ± 4.4 51.2 ± 6.0 57.6 ± 7.2

44.2 ± 4.1 54.0 ± 5.7 60.7 ± 6.9

46.8 ± 3.9 57.2 ± 5.5 64.3 ± 6.7

48.3 ± 3.9 59.0 ± 5.5 66.3 ± 6.7

50.1 ± 3.9 61.2 ± 5.5 68.8 ± 6.7

51.2 ± 4.0 62.6 ± 5.6 70.3 ± 6.8

52.7 ± 4.1 64.4 ± 5.8 72.4 ± 7.0

DMLQED2′ (Gy) 56.0 ± 1.0 58.0 ± 1.1 60.0 ± 1.1

59.1 61.2 63.3

62.5 ± 1.2 64.8 ± 1.2 67.0 ± 1.3

64.5 ± 1.9 66.8 ± 1.9 69.0 ± 2.0

67.0 ± 2.7 69.3 ± 2.8 71.7 ± 2.9

72.5 ± 3.2 75.1 ± 3.4 77.6 ± 3.5

70.5 ± 3.9 73.0 ± 4.1 75.5 ± 4.2

(A) 3.3 cc 1 cc 0.21 cc (B) 5 cc 2 cc Max dose

The variation of parameters found in Murphy et al11 (A) and Verma et al13 (B) at levels of 5%, 7%, 10%, 12%, 15%, 17%, and 20%. Error bars represent the standard error associated with calculations of DMLQED2′. The Verma et al data at NTCP = 7% are originally from the abstract, where only mean dose values were reported and therefore error bars were not estimated.

Practical Radiation Oncology: March-April 2014

In conclusion, parameters for the MLQ-based LKB NTCP model were found that allow for the calculation of concordant isoeffective dose–response data of duodenum toxicity. The (DMLQED2',v) can be used to determine appropriate duodenum and small bowel constraints for treatment planning. Though these model parameters can be used in RT pancreatic dose escalation planning, strategies accounting to organ motion and day-to-day variability, such as gated IGRT, will be essential in decreasing tumor margins in the pancreatic head and decreasing duodenum dose.

Duodenal and small bowel toxicity data

10.

11.

12.

13.

14.

References 1. Gastrointestinal Tumor Study Group. Treatment of locally unresectable carcinoma of the pancreas: comparison of combined-modality therapy (chemotherapy plus radiotherapy) to chemotherapy alone. J Natl Cancer Inst. 1998;80:751-755. 2. Klaassen DJ, MacIntyre JM, Catton GE, Engstrom PF, Moertel CG. Treatment of locally unresectable cancer of the stomach and pancreas: a randomized comparison of 5-fluorouracil alone with radiation plus concurrent and maintenance 5-fluorouracil—an Eastern Cooperative Oncology Group study. J Clin Oncol. 1985;3:373-378. 3. Loehrer PT, Powell ME, Cardenes HR, et al. A randomized phase III study of gemcitabine in combination with radiation therapy versus gemcitabine alone in patients with localized, unresectable pancreatic cancer: E4201. J Clin Oncol. 2008;26(Suppl):S214. [abstract 4506]. 4. Ceha MH, van Tienhoven G, Gouma DJ, et al. Feasibility and efficacy of high dose conformal radiotherapy for patients with locally advanced pancreatic carcinoma. Cancer. 2000;89:2222-2229. 5. Landry JC, Yang GY, Ting JY, et al. Treatment of pancreatic cancer tumors with intensity-modulated radiation therapy (IMRT) using the volume at risk approach (VARA): employing dose-volume histogram (DVH) and normal tissue complication probability (NTCP) to evaluate small bowel toxicity. Med Dosim. 2002;27:121-129. 6. Li XA, Liu F, Tai A, et al. Development of an online adaptive solution to account for inter- and intra-fractional variations. Radiother Oncol. 2011;100:370-374. 7. Kirkpatrick JP, Brenner DJ, Orton CG. Point/counterpoint. The linear-quadratic model is inappropriate to model high dose per fraction effects in radiosurgery. Med Phys. 2009;36:3381-3384. 8. Guerrero M, Li XA. Extending the linear-quadratic model for large fraction doses pertinent to stereotatic radiotherapy. Phys Med Biol. 2004;49:4825-4835. 9. Comuzzi C, Polese P, Melchior A, Portanova R, Tolazzi M. SOLVERSTAT: a new utility for multipurpose analysis. An

15.

16.

17.

18.

19.

20. 21.

22.

23.

24.

25.

e131

application to the investigation of deoxygenated Co(II) complex formulation in dimethylsulfoxide solution. Talanta. 2003;59:67-80. Burman C, Kutcher GJ, Emami B, Goitein M. Fitting of normal tissue tolerance data to an analytic function. Int J Radiat Oncol Biol Phys. 1991;21:123-135. Murphy JD, Christman-Skieller C, Kim J, Dieterich S, Chang DT, Koong AC. A dosimetric model of duodenal toxicity after stereotactic body radiotherapy for pancreatic cancer. Int J Radiat Oncol Biol Phys. 2010;78:1420-1426. Kelly P, Das P, Pinnix CC, et al. Duodenal toxicity after fractionated chemoradiation for unresectable pancreatic cancer. Int J Radiat Oncol Biol Phys. 2013;85:e143-e149. Verma J, Klopp AH, Jhingran A, et al. Dosimetric predictors of duodenal toxicity following IMRT to para-aortic nodes. Int J Radiat Oncol Biol Phys. 2011;81:S473. Kopek N, Holt MI, Hansen AT, Høyer M. Stereotactic body radiotherapy for unresectable cholangiocarcinoma. Radiother Oncol. 2010;94:47-52. Deore SM, Shrivastava SK, Supe SJ, Viswanathan PS, Dinshaw KA. α/β value and importance of dose per fraction for the late rectal and recto-sigmoid complications. Strahlenther Onkol. 1993;169:521-526. Dische S, Saunders MI, Sealy R, et al. Carcinoma of the cervix and the use of hyperbaric oxygen with radiotherapy: a report of a randomized controlled trial. Radiother Oncol. 1999;53:93-98. Bentzen SM, Joiner MC. The linear-quadratic approach in clinical practice. In: Joiner M, van der Kogel A, eds. Basic clinical radiobiology. 4th ed. London: Edward Arnold Press; 2009120-134. Schellenberg D, Goodman KA, Lee F, et al. Gemcitabine chemotherapy and single-fraction stereotactic body radiotherapy for locally advanced pancreatic cancer. Int J Radiat Oncol Biol Phys. 2008;72:678-686. Emami B, Lyman J, Brown A, et al. Tolerance of normal tissue to therapeutic irradiation. Int J Radiat Oncol Biol Phys. 1991;21: 109-122. Potish RA, Jones Jr TK, Levitt SH. Factors predisposing to radiationrelated small-bowel damage. Radiology. 1979;132:479-482. Gallagher MJ, Brereton HD, Rostock RA, et al. A prospective study of treatment techniques to minimize the volume of pelvic small bowel with reduction of acute and late effects associated with pelvic irradiation. Int J Radiat Oncol Biol Phys. 1986;12:1565-1573. Lillemoe KD, Brigham RA, Harmon JW, Feaster MM, Saunders JR, d'Avis JA. Surgical management of small-bowel radiation enteritis. Arch Surg. 1983;118:905-907. King CR, Lehmann J, Adler JR, Hai J. Cyberknife radiotherapy for localized prostate cancer: rationale and technical feasibility. Technol Cancer Res Treat. 2003;2:25-30. Bentzen SM, Vaeth M, Pedersen DE, Overgaard J. Why actuarial estimates should be used in reporting late normal-tissue effects of cancer treatment … now! Int J Radiat Oncol Biol Phys. 1995;32:1531-1534. Ten Haken RK, Martel MK, Kessler ML, et al. Use of Veff and isoNTCP in the implementation of dose escalation protocols. Int J Radiat Oncol Biol Phys. 1993;27:689-695.