Computed Tomography for Head-and-Neck Cancer Treatment Planning Necessary?

International Journal of

Radiation Oncology biology

physics

www.redjournal.org

Clinical Investigation: Head and Neck Cancer

Is Image Registration of FluorodeoxyglucoseePositron Emission Tomography/Computed Tomography for Head-and-Neck Cancer Treatment Planning Necessary? David Fried, B.S.,* Michael Lawrence, Ph.D.,* Amir H. Khandani, M.D.,y Julian Rosenman, M.D., Ph.D.,*,z Tim Cullip, M.S.,* and Bhishamjit S. Chera, M.D.*,z *Department of Radiation Oncology, yDepartment of Radiology, and zLineberger Comprehensive Cancer Center, University of North Carolina at Chapel Hill, Chapel Hill, NC Received Aug 26, 2011, and in revised form Oct 26, 2011. Accepted for publication Dec 20, 2011

Summary We evaluated the dosimetry and patterns of failure related to fluorodeoxyglucoseePET edefined biological tumor volumes (BTVs) for headand-neck squamous cell carcinoma (HNSCC) treated with definitive radiotherapy. BTVs and CTedefined volumes had less than complete overlap. BTVs were smaller than CT-defined targets. Dosimetric coverage was similar between failed and controlled groups, as well as between BTVs and CT defined volumes. CT-based delineation alone may be sufficient for treatment planning in patients with HNSCC.

Purpose: To evaluate dosimetry and patterns of failure related to fluorodeoxyglucoseepositron emission tomography (FDG-PET)edefined biological tumor volumes (BTVs) for head-and-neck squamous cell carcinoma (HNSCC) treated with definitive radiotherapy (RT). Methods and Materials: We conducted a retrospective study of 91 HNSCC patients who received pretreatment PET/CT scans that were not formally used for target delineation. The median follow-up was 34.5 months. Image registration was performed for PET, planning CT, and post-RT failure CT scans. Previously defined primary (CTPRIMARY) and nodal (CTNODE) gross tumor volumes (GTV) were used. The primary BTV (BTVPRIMARY) and nodal BTV (BTVNODE) were defined visually (PETvis). The BTVPRIMARY was also contoured using 40% and 50% peak PET activity (PET40, PET50). The recurrent GTVs were contoured on post-RT CT scans. Dosimetry was evaluated on the planning-CT and pretreatment PET scan. PET and CT dosimetric/volumetric data was compared for those with and without local-regional failure (LRF). Results: In all, 29 of 91 (32%) patients experienced LRF: 10 local alone, 7 regional alone, and 12 local and regional. BTVs and CT volumes had less than complete overlap. BTVs were smaller than CT-defined targets. Dosimetric coverage was similar between failed and controlled groups as well as between BTVs and CT-defined volumes. Conclusions: PET and CT-defined tumor volumes received similar RT doses despite having less than complete overlap and the inaccuracies of image registration. LRF correlated with both CT and PET-defined volumes. The dosimetry for PET- and/or CT-based tumor volumes was not significantly inferior in patients with LRF. CT-based delineation alone may be sufficient for treatment planning in patients with HNSCC. Image registration of FDG-PET may not be necessary. Ó 2012 Elsevier Inc. Keywords: PET, Head and neck cancer, Radiotherapy

Reprint requests to: Bhishamjit S. Chera, M.D., 101 Manning Drive CB #7512, Chapel Hill, NC 27514. Tel: 919-445-5286; Fax: 919-9667681; E-mail: [email protected] Int J Radiation Oncol Biol Phys, Vol. 84, No. 3, pp. 748e754, 2012 0360-3016/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.ijrobp.2011.12.071

Conflict of interest: none.

Volume 84  Number 3  2012

Registration of FDG-PET/CT for head-and-neck treatment planning

Introduction 18

F-fluorodeoxyglucoseepositron emission tomography (FDGPET) has become a widely used biological imaging technique for many cancers. For head-and-neck squamous cell carcinoma (HNSCC) patients, multiple publications have concluded that FDG-PET is useful for identifying gross disease, for staging purposes, and for target delineation for radiotherapy (RT) (1e7). It is suggested that FDG-PETebased volumes may be more accurate and reproducible than computed tomography (CT) target delineation. However, the usefulness of FDG-PET in RT planning is not straightforward, as there is no consensus for proper delineation methods. In addition, FDG-PET co-registration requires added time/effort from physicists and physicians. Physicists need to identify, retrieve, and register the FDG-PET scans for RT planning, and physicians need to contour additional targets based on the registered FDG-PET. Regardless of delineation method, it has been observed that CT-based volumes and BTVs may not be correlated with one another. Schinagl et al. reported that at least 20% of BTVs were outside the CT-based volume in approximately 30% to 65% of patients, and that the average percent overlap of the BTVs and CT-based volumes was 66% to 85% (8). Paulino et al. conducted a retrospective study of 40 HNSCC patients who underwent IMRT using CT-based targets. A 50% threshold technique was used to delineate the primary tumor, and these authors concluded that the volume receiving at least 95% of the prescribed dose was suboptimal in 25% of patients (9). Furthermore, a study correlating imaging with histopathologic findings found evidence that both CT and FDG-PET overestimate the extent of tumor (10). Despite discordance between CT-based volumes and BTVs, the majority of failures occur “in-field” (11). Although FDG-PET is purported to be useful for target delineation, it remains unclear whether incorporation of BTVs into CT-based treatment planning alters dosimetry or local-regional control. We hypothesize that the dose to the pre-treatment BTVs is adequate even when not accounted for in treatment planning. We conducted a retrospective analysis of dose delivered to the pre-treatment BTV in HNSCC patients who had CT-based treatment planning without FDG-PET target delineation. A patterns of failure analysis was also performed correlating local-regional failures with BTV and CT-based gross tumor volume (GTV).

Methods and Materials Patient population We obtained approval from our Institutional Review Board for this study (IRB # 09-2146), and conducted a retrospective study of 91 HNSCC patients receiving pretreatment FDG-PET/CT scans followed by definitive RT between April 2003 and January 2010. Patients were identified using the Lineberger Comprehensive Cancer Center Head and Neck Database.

Pretreatment evaluation and assessment of HNSCC patients From 2003 to 2010, HNSCC patients received thorough physical examinations by an otolaryngologist, radiation oncologist, and

749

medical oncologist that included fiberoptic laryngoscopy, examination under anesthesia with panendoscopy, and contrastenhanced CT scans of the neck and chest. If FDG-PET was performed, it was obtained after the aforementioned evaluation. FDG-PET in combination with a CT scan (FDG-PET/CT) was implemented in January 2003. FDG-PET imaging was performed on a PET/CT scanner (single-slice or 40-slice Biograph, Siemens, Knoxville, TN). Patients were discussed to determine staging and treatment recommendations at our Head and Neck Multidisciplinary Tumor Board. Restaging was done according to the sixth edition of the American Joint Committee on Cancer staging guidelines (12).

Definitive chemoradiotherapy Patients were treated with three-dimensional conformal radiotherapy (3D-CRT) or intensity-modulated radiotherapy (IMRT). CT-based targets were predominantly used to determine treatment volumes. IMRT was commonly used when the contralateral parotid could be spared without compromising target coverage. Standard and high-risk target volumes and individual seven- to nine-field IMRT plans were created for each target. The 3D-CRT plans consisted of a three-field technique, i.e., opposed laterals matched to a low anterior neck field. The majority of patients (93%) received concurrent chemotherapy (Table 1). Patients had post-RT imaging (e.g., CT or FDG-PET) to assess for nodal disease and necessity of neck dissection. All patients were followed clinically and radiographically by a head-and-neck surgeon and a radiation oncologist.

Image registration and target delineation The CT-based RT plans were unarchived from our in-house treatment planning system (PlanUNC, PLUNC). Historically, at our institution pretreatment PET scans were not routinely used in the treatment planning but, rather, in an initial diagnostic staging evaluation. Radiation oncologists used the pre-treatment PET scans to identify the location of the primary tumor, to identify positive nodal levels within the neck, and to exclude the presence of metastases and/or second primary sites. No target volumes were contoured (visually or quantitatively) on the PET/CT for treatment planning purposes. Accurate PET/CT registration is of the utmost importance for target delineation. For this study, image registration was retrospectively performed for diagnostic PET/CT and planning CT. PET/CT were not done in the treatment position with appropriate immobilization, and the average time from PET/CT scan to CT-simulation scan was 22 days. Thus, registration errors may be more pronounced in this study cohort. We used a rigid body registration methodology (described below) as described by Hwang et al. (13). PET scans were registered in PLUNC using their companion pre-registered CT scan. PLUNC allows for modifying secondary images for translation and rotation in the axial, sagittal, and coronal planes. The companion CT was registered to the closest anatomical landmark to the tumor location on the planning CT (e.g. vertebral body, mandible, or cricoid cartilage) without any GTV contour visible. Previously defined primary (CTPRIMARY) and nodal (CTNODE) GTVs from the planning CT were used. CT criteria for nodal positivity included the appearance of central necrosis, enhancement, or a size of greater than 10 mm. The primary BTV (BTVPRIMARY) and nodal BTV (BTVNODE) were defined via visual interpretation (PETvis). PETvis was defined as the edge which the PET avid volume is considered elevated compared with the background. PETvis is inherently

750

International Journal of Radiation Oncology  Biology  Physics

Fried et al.

Table 1

Patient demographic and treatment characteristics

Median age (y) Median prescribed dose Gender Male Female Primary site Hypopharynx Larynx Nasopharynx Oral cavity Oropharynx Unknown Tumor stage T0 T1 T2 T3 T4 Nodal stage N0 N1 N2a N2b N2c N3 Overall stage II III IV RT type 3DCRT IMRT Platinum chemo Yes No

All patients

Controlled

n

%

n

91

Failed %

n

%

100

62

67

29

23

57 7,000

NA

57 7,000 cGy (6,600e7,800)

NA NA

56 7,000 cGy (6,500e7,800)

NA NA

70 21

77 23

46 16

74 26

24 5

83 17

9 15 5 7 50 5

10 15 6 8 55 6

6 10 4 2 37 3

10 16 6 3 60 5

3 5 1 5 13 2

10 17 3 17 45 7

5 7 24 25 30

6 8 26 27 33

3 7 18 20 14

5 11 29 32 23

2 0 6 5 16

7 0 21 17 55

12 9 4 20 27 19

13 10 4 22 30 21

9 7 2 16 17 11

15 11 3 26 27 18

3 2 2 4 10 8

10 7 7 14 34 28

3 12 76

3 13 84

1 10 51

2 16 82

2 2 25

7 7 86

23 67

25 75

11 51

18 82

13 16

45 55

84 7

92 8

58 4

94 6

26 3

90 10

Abbreviations: RT Z radiation therapy; 3DCRT Z three-dimensional conformal radiotherapy; IMRT Z intensity-modulated radiotherapy.

subjective but was included for comparison, as it is the common method used by radiation oncologist. Nuclear medicine physicians did not set margins or threshold levels for radiation oncologists when determining PETvis contours. Nuclear medicine physicians were not consulted when contouring PETvis as this is not routine clinical practice in radiation oncology. The BTVPRIMARY was also defined by the contour of 40% and 50% of the peak PET activity (PET40, PET50). Patients with recurrence had their recurrent primary and nodal GTVs contoured on a post-RT CT scans. The overlap fraction (OF) of BTV (OFPET) and CTGTV (OFCT) was calculated by dividing the overlap portion of the PET and CT-defined volumes by the entire BTV or CT volume, respectively (10).

Evaluation of dose delivered to BTVs and CT-based GTVs RT dose was evaluated on the planning-CT and respective secondary image sets. Dosimetry was compared for BTV and CT

volumes and for those with and without LRF. The prescribed dose, dose to 100% volume, and dose to 95% volume were collected for the target volumes. We evaluated whether 95% of the prescribed dose covered 95% of the volume and whether the minimum dose within 95% of volume was greater than 95% of the prescription dose.

Patterns of failure The post-treatment diagnostic CT for patients with a localregional failure was registered to their corresponding pretreatment scans. Of 29 LRF, 24 were biopsy proved. Four of 6 nonbiopsied recurrences were post distant failures. In the remaining 2 nonbiopsied patients, the presence of recurrence was clearly evident and was not biopsied. The average time between treatment termination and acquisition of the failure CT was 190 days. Of the 91 patients, 64 (70%) received PET/CT scans post

Abbreviations: BTV Z biological tumor volume; CT Z computed tomography; OF Z overlap fraction; PET Z fluorodeoxyglucoseepositron emission tomography.

0.39 (0.05e0.71) NA 0.56 (0.16e0.88) NA 16.1 (0.8e95.9) 21.5 (2e125.3) 0.14 (0e0.91) NA 0.50 (0e1) NA 6.8 (0.5e90.3) 15.6 (0.7e187) 0.19 (0e0.91) NA 8 (0.5e95.9) 16 (0.7e187) BTVNODE CTNODE

0.52 (0e1) NA

Median OFCT Median OFPET Median OFPET Median (cc) Median OFCT Median (cc)

Median OFPET

Controlled regionally (n Z 72) All patients (n Z 91)

Median OFCT

Median (cc)

Failed regionally (n Z19)

0.59 (0.9e0.89) 0.68 (0.12e0.94) 0.77 (0.16e1) NA (6.3e185) (4e110) (1.8e73) (14.6e240) 40.4 19.1 12.6 50.3 0.28 (0e0.86) 0.20 (0e0.72) 0.12 (0e0.63) NA 0.52 (0e1) 0.65 (0e1) 0.73 (0e1) NA (0.6e100) (0.5e71) (0.1e58) (1.5e179) 15.4 9.7 6.1 25.6 0.32 (0e0.86) 0.21 (0e0.72) 0.13 (0e0.63) NA 0.55 (0e1) 0.65 (0e1) 0.74 (0e1) NA

Median OFPET Median (cc) Median OFCT Median OFPET

All patients (n Z 91)

Delineation method for regional tumor

Of the 91 patients, 29 (32%) patients developed LRF: 10 (11%) isolated local, 7 (8%) isolated regional, 12 (13%) local and

Delineation method for primary tumor

Patterns of failure

Volumetric information for various target delineation methods

Dosimetric data are shown in Table 3. BTVs and CTGTV received similar RT doses for the primary and regional tumor sites despite incomplete overlap and inherent inaccuracies of image registration. It should be noted that 95% of the prescription dose did not cover 95% of the primary GTV for 29% of the PETVIS, 19% of the PET40, 15% of the PET50, and 12% of the CTPRIMARY. The minimum dose in 95% of the primary GTV was less than 95% of the prescription dose for 24% of the PETVIS, 19% of the PET40, 9% of the PET50, and 11% of the CTPRIMARY. There were negligible differences between the dose to the CT and BTV targets for patients with local control vs. those with local failure (0.2e2.8 Gy). We also did not observe any substantial differences for the dose received to 95% volume between smaller (T1/T2) and larger (T3/T4) primary tumors or smaller (N1eN2b) and larger (N2ceN3) nodal disease. There were also no substantial differences in dose received by 95% between low volume (31.1 cc) PET-avid disease vs. high-volume (>31.1 cc). The difference in dose between patients with and without regional failure ranged from 2.3 Gy to 7.8 Gy.

Table 2

Dosimetric data

Controlled locally (n Z 69)

Table 2 shows the median tumor volumes and OF for the primary and regional sites. The median BTVPRIMARY defined via PET40 and PET50 were similar in size but significantly smaller than PETVIS. The CTPRIMARY and PETVIS were similar in size for the primary tumor. The OFPET and OFCT for the primary tumor ranged from 0.55 to 0.74 and 0.13 to 0.32, respectively. Patients with a LRF had BTVs that were significantly larger than those without a LRF. Furthermore, the OFPET was similar between the failed and controlled groups. OFPET and OFCT values were slightly larger (w0.1e0.2) in larynx/hypopharynx vs. oropharynx/oral cavity primaries. The average standardized uptake value (SUV) within the overlap of BTVPRIMARY and CTPRIMARY was 7.3. This was approximately two times higher than the uptake within the nonoverlapping areas of the BTVPRIMARY and/or CTPRIMARY alone (SUV Z 3.7). The BTVNODE was smaller than the CTNODE, and the OFPET and OFCT were 0.52 and 0.19 for patients with nodal disease. Similar to the primary tumor, the nodal target volumes were larger in patients with regional recurrence vs. those with regional control, and the OFPET and OFCT were larger in patients with regional recurrence.

Median OFCT

Median (cc)

GTV data

(0.6e185) (0.5e110) (0.1e73) (1.5e240)

The median follow-up in living patients was 35 months (range, 5e93 months). Demographic and staging information is shown in Table 1.

29 10.8 7 32

Failed locally (n Z 22)

Patient characteristics

BTV PETVIS PET40 PET50 CTPRIMARY

Results

Median OFPET

Median OFCT

RT. The recurrent/persistent disease was contoured and correlated with the BTV and planning CTGTV.

0.4 (0.05e0.76) 0.24 (0.04e0.57) 0.17 (0.03e0.47) NA

Registration of FDG-PET/CT for head-and-neck treatment planning

Median (cc)

Volume 84  Number 3  2012

751

International Journal of Radiation Oncology  Biology  Physics regional. Of these patients, 27 had radiographic confirmed LRF and were included in the patterns of failure analysis. A total of 18 of 22 local failures and 18 of 19 nodal failures were observed on surveillance CT scans. The patterns of LRF seen radiographically are shown in Table 4, and SUV data regarding LRF is in Table 5. Both CT and BTV volumes had a high correlation with local and regional failures. One primary failure was outside of all BTVs but within the CTPRIMARY. Recurrence was in the base of tongue but initially delineated on PET in the retromolar trigone region. The CTPRIMARY encompassed the retromolar trigone and base of tongue. One contralateral nodal failure was observed that was not delineated by either method. Finally, PET failed to detect a 1-cm lymph node that ultimately failed. Overall, 90% to 95% of LRFs occurred within the BTV or CT-based target volumes. The maximum SUV and average SUV in PET40 and PET50 were statistically different (p < 0.05), and the average SUV in PETVIS was marginally different (p Z 0.06) between those patients with and without LRF. Abbreviations: BTV Z biological tumor volume; CT Z computed tomography; PET Z fluorodeoxyglucoseepositron emission tomography.

67.4 Gy (27.5e72.5) 68.6 Gy (27.5e72.5) 60.8 Gy (39.5e71.7) 7.8 Gy 68.9 Gy (46.5e72.7) 69.1 Gy (46.5e72.7) 66.8 Gy (55.9e72.9) 2.3 Gy 68.3 Gy (35.6e72.9) 68.4 Gy (51.8e72.9) 67.9 Gy (34.6e72.0) 0.5 Gy 68.6 Gy (57.6e72.7) 68.7 Gy (57.6e71.6) 68.5 Gy (58.6e72.7) 0.2 Gy

68.8 Gy (52.2e73.1) 68.9 Gy (52.2e73.1) 68.3 Gy (53.3e72.7) 0.5 Gy

68.9 Gy (53.2e73.3) 69.1 Gy (53.2e73.3) 68.7 Gy (63.1e73.0) 0.4 Gy

59.7 Gy (15.4e71.7) 61.2 Gy (15.4e71.7) 57.6 Gy (23.9e68.8) 3.6 Gy 65.1 Gy (28.4e72.4) 66.2 Gy (28.4e72.4) 60.9 Gy (41.1e69.0) 5.3 Gy 67.9 Gy (24.4e72.4) 67.9 Gy (41.5e72.4) 67.5 Gy (24.4e72.2) 0.4 Gy 65.1 Gy (16.7e71.1) 65.4 Gy (30.6e71.1) 62.6 Gy (16.7e69.4) 2.8 Gy

Median dose to 100% volume All patients (n Z 91) Controlled (n Z 62) Failed (n Z 29) Difference between failed and controlled Median dose to 95% volume All patients Controlled Failed Difference between failed and controlled

66.5 Gy (14.6e70.6) 67.0 Gy (14.6e70.6) 65.1 Gy (43.0e70.5) 1.9 Gy

66.7 Gy (17.4e72.4) 67.4 Gy (38.9e72.4) 66.2 Gy (17.4e72.2) 1.2 Gy

CTNODE BTVPRIMARY (PETvis) CTPRIMARY

Dose delivered to the FDG-PET and CT tumor volumes

BTVPRIMARY (PET40)

BTVPRIMARY (PET50)

BTVNODE (PET)

Fried et al.

Table 3

752

Discussion We retrospectively evaluated the radiation dose to pre-treatment FDG-PET delineated BTVs in HNSCC patients receiving CTbased RT planning. The pre-treatment FDG-PET scan was not registered for treatment planning in this cohort. Furthermore, we assessed the patterns of LRF in relation to BTVs. BTVs were smaller than the CT-based targets. This was also true for patients with and without local-regional control. Patients with LRF had larger BTV and CT-based volumes. We observed discordance in overlap between BTVs and CT-based volumes, which complement previous observations (8). Despite dose being prescribed to CT-based volumes, discrepancy in overlap, and inaccuracies of image registration, the dose received by the BTVs were similar to that received by the CT targets (Fig.). The dose to BTVs and CT-based volumes, and their relative overlap fractions, were similar between patients with and without LRF. Higher uptake was observed in the overlap between BTVPRIMARY and CTPRIMARY, and most LRF occurred within both the BTVs and the CT-based target volumes. Thus, even though the pre-treatment FDG-PET was not formally registered a priori to the CT simulation scan for treatment planning purposes, the dose received by BTVs and the patterns of failure were comparable. Several publications evaluated SUV threshold-based (e.g., SUVmax, SUV50%, SUV75%) target delineation methods and discordance between the CT-based targets and PET-based target have been observed (5, 8, 14). Some investigators have concluded that FDG-PET based delineation is more accurate; however, there is no consensus as to which SUV threshold most accurately delineates malignant tissue from benign tissue. Our observations are comparable to those in previous FDGPET HNSCC reports. Schinagl et al. reported larger BTVPRIMARY for PETVIS, PET40, and PET50 and observed OFPET of 0.66, 0.72, and 0.80, respectively (8). We observed smaller BTVs and an OFPET similar to those observed by Schinagal et al. (PETVIS, 0.55; PET40, 0.66; and PET50, 0.74). However, their OFCT was two to three times higher than our cohort for all delineation methods. This could be due to differences in CT contouring between physicians, registration error, or differences in FDG-PET protocol. We observed slightly improved dosimetric coverage in comparison to a study of 40 HNSCC patients conducted by Paulino et al., in which the investigators study observed that 25%

Volume 84  Number 3  2012 Table 4

Registration of FDG-PET/CT for head-and-neck treatment planning

753

Patterns of local and regional failure

Local failures Neck failures

CTPRIMARY/Node

BTVPRIMARY/Node (PETvis)

BTVPRIMARY (PET40)

BTVPRIMARY (PET50)

18/18 17/18

17/18 16/18

17/18 NA

17/18 NA

Abbreviations: BTV Z biological tumor volume; CT Z computed tomography; PET Z fluorodeoxyglucoseepositron emission tomography.

of the PET50 volumes had suboptimal coverage of 95% of the prescribed dose and the minimum dose within 95% of PET50. Our study had 15% PET50 volumes that did not meet the 95% dose coverage constraint and 9% that did not achieve the minimum dose constraint. Numerous investigators have suggested that FDG-PET is beneficial when contouring for RT treatment planning. FDG-PET has been shown to reduce interobserver variability in HNSCC in a study by Ciernik et al., which showed reduction of the difference in mean volume from 25.7 cc with CT-based targets to 9.2 cc with BTV (7). Daisne et al. concluded that in pharyngolarngeal cancers, PET is superior to CT and MRI for accurately depicting GTV when compared with surgical specimens. This reduction in GTV could possibly enable further normal tissue sparing and dose escalation. In our study, higher SUV values were observed in the overlap region between CT-based volumes and BTVs. Furthermore, 17 of 18 local failures were found to be in PET50. Significantly higher maximum and average SUV was seen for patients with LRF compared with those controlled. One could conclude that higher tumor metabolism may indicate radiation resistance and thusly a need for higher dose. It has been implied that dose escalation to areas of high FDG uptake could reduce the rate of local failure (15e18). Dose escalation has been studied in prospective trials using voxel-based/subvolume methods. It has been shown that dose escalation to FDG volumes is feasible and well tolerated, and that dosimetric differences outside the escalated volumes are minimal (18). However, no data are available that illustrate a control benefit or toxicity reduction due to smaller treatment volumes. In a prospective study by Mandani et al., more than half of the local failures were observed in the FDG uptake region that received the highest dose level (17). Further study of “dose painting” and/or adaptive treatment planning strategies based on BTVs with the goal of increasing tumor control and reducing toxicity are warranted. This study is, to our knowledge, the first to compare PET/CT volumetric and dosimetric data between patients with and without LRF. We observed similar values for OFPET and OFCT, and dosimetric coverage of all BTVs and CT tumor volumes were similar

Table 5

between those failed and controlled (differing by only 1e2 Gy on average). The only significant difference found between failed and controlled groups was in terms of tumor volume, mean SUV, and maximum SUV. Thus, in this HNSCC patient cohort, the cause of LRF may not be inadequate dosimetric coverage but may be more dependent on tumor burden and aggressive biology. Even though formal use of FDG-PET scans may not be necessary for GTV delineation, FDG-PET scans may be beneficial or useful in other ways. FDG-PET scans are known to be useful for staging and for the management of the post-RT node-positive neck. Numerous other studies have observed pre- and posttreatment quantitative assessments of FDG uptake (e.g., metabolic tumor volume, maximum SUV, mean SUV) to be prognostic for clinical outcomes. FDG-PET scans may be used for risk stratification, and treatments may be tailored to their estimated risk of recurrence. Furthermore, the addition of PET scans could also serve as a valuable quality assurance tool by confirming the definition of CT volumes to reduce gross errors (e.g., right vs. left). The prevailing opinion is that registration of FDG-PET scans to CT planning appears to be beneficial. The purported main benefit is that FDG-PET more accurately delineates the target volume. There is no substantial evidence proving FDGPETebased target delineation to be superior to CT-based target delineation for HNSCC treatment planning. Some institutions have adopted treating both PET and CT defined volumes to prescription dose (9). However, this may lead to larger target volumes and thus higher normal tissue dose. We did not observe inadequate dosimetric coverage of the PET defined BTVs in our institution specific patients for whom PET was not used a priori for treatment planning. Specifically including PET-defined volumes may not significantly improve local-regional control, based on our findings, and thus image registration of FDG-PET may not be necessary. Prioritizing dosimetric coverage of the CT-based targets will indirectly result in adequate dose to the BTVs.

Association of SUV with local-regional failure Local-regional Local-regional p failure control value

Average SUV PETVIS Max SUV PETVIS Average SUV PET40 Average SUV PET50

7.3 16.7 9.8 11.2

6.6 13.3 7.9 8.9

.06 .02 .02 .02

Abbreviations: Max Z maximum; PET Z fluorodeoxyglucoseepositron emission tomography; SUV Z standardized uptake value.

Fig. T3N2cM0 of the supraglottic larynx. Biological tumor volumes (BTV) (PETVIS) are represented by a dark grey solid line. CT-based tumor is indicated by black solid line. The misregistration is most notable for the nodal disease. The 100% isodose line (IDL) is represented by the dashed white line, and the 95% IDL by the dashed black line. Despite overlap discordance, the 95% IDL encompasses BTV and CT-based volumes.

754

Fried et al.

Conclusions In this study, PET and planning CT-defined tumor volumes received similar RT doses for both primary and nodal tumor sites despite having less than complete overlap and the inherent inaccuracies of image registration. LRF correlated with both CT and PET defined volumes. Dosimetry for PET and/or planning CT-defined tumor volumes were not significantly inferior in patients with LRF. CTbased target delineation alone may be sufficient for treatment planning in patients with HNSCC. Image registration of FDG-PET for treatment planning purposes may not be necessary for gross tumor delineation; however, it may be useful in prognostication and in future efforts to individualize radiation treatments.

References 1. Lonneux M, Hamoir M, Reychler H, et al. Positron emission tomography with [18F]fluorodeoxyglucose improves staging and patient management in patients with head and neck squamous cell carcinoma: A multicenter prospective study. J Clin Oncol 2010;28:1190e1195. 2. Guido A, Fuccio L, Rombi B, et al. Combined 18F-FDG-PET/CT imaging in radiotherapy target delineation for head-and-neck cancer. Int J Radiat Oncol Biol Phys 2009;73:759e763. 3. Ng SH, Yen TC, Liao CT, et al. 18F-FDG PET and CT/MRI in oral cavity squamous cell carcinoma: A prospective study of 124 patients with histologic correlation. J Nucl Med 2005;46:1136e1143. 4. Schwartz DL, Ford E, Rajendran J, et al. FDG-PET/CT imaging for preradiotherapy staging of head-and-neck squamous cell carcinoma. Int J Radiat Oncol Biol Phys 2005;61:129e136. 5. Murakami R, Uozumi H, Hirai T, et al. Impact of FDG-PET/CT imaging on nodal staging for head-and-neck squamous cell carcinoma. Int J Radiat Oncol Biol Phys 2007;68:377e382. 6. Gregoire V, Haustermans K, Geets X, et al. PET-based treatment planning in radiotherapy: A new standard? J Nucl Med 2007;48(Suppl 1):68Se77S. 7. Ciernik IF, Dizendorf E, Baumert BG, et al. Radiation treatment planning with an integrated positron emission and computer tomography (PET/CT): A feasibility study. Int J Radiat Oncol Biol Phys 2003;57:853e863.

International Journal of Radiation Oncology  Biology  Physics 8. Schinagl DA, Vogel WV, Hoffmann AL, et al. Comparison of five segmentation tools for 18F-fluoro-deoxy-glucose-positron emission tomography-based target volume definition in head and neck cancer. Int J Radiat Oncol Biol Phys 2007;69:1282e1289. 9. Paulino AC, Koshy M, Howell R, et al. Comparison of CT- and FDGPET-defined gross tumor volume in intensity-modulated radiotherapy for head-and-neck cancer. Int J Radiat Oncol Biol Phys 2005;61: 1385e1392. 10. Daisne JF, Duprez T, Weynand B, et al. Tumor volume in pharyngolaryngeal squamous cell carcinoma: Comparison at CT, MR imaging, and FDG PET and validation with surgical specimen. Radiology 2004;233:93e100. 11. Soto DE, Kessler ML, Piert M, et al. Correlation between pretreatment FDG-PET biological target volume and anatomical location of failure after radiation therapy for head and neck cancers. Radiother Oncol 2008;89:13e18. 12. Greene FL, Page DL, Fleming ID, et al., American Joint Committee on Cancer, American Cancer Society. AJCC cancer staging manual. 6th ed. New York: Springer-Verlag; 2002. 13. Hwang AB, Bacharach SL, Yom SS, et al. Can positron emission tomography (PET) or PET/computed tomography (CT) acquired in a nontreatment position be accurately registered to a head-and-neck radiotherapy planning CT? Int J Radiat Oncol Biol Phys 2009;73: 578e584. 14. Greco C, Nehmeh SA, Schoder H, et al. Evaluation of different methods of 18F-FDG-PET target volume delineation in the radiotherapy of head and neck cancer. Am J Clin Oncol 2008;31: 439e445. 15. Duprez F, De Neve W, De Gersem W, et al. Adaptive dose painting by numbers for head-and-neck cancer. Int J Radiat Oncol Biol Phys 2011; 80:1045e1055. 16. Schwartz DL, Ford EC, Rajendran J, et al. FDG-PET/CT-guided intensity modulated head and neck radiotherapy: A pilot investigation. Head Neck 2005;27:478e487. 17. Madani I, Duthoy W, Derie C, et al. Positron emission tomographyguided, focal-dose escalation using intensity-modulated radiotherapy for head and neck cancer. Int J Radiat Oncol Biol Phys 2007;68: 126e135. 18. Vanderstraeten B, Duthoy W, De Gersem W, et al. [18F]fluoro-deoxyglucose positron emission tomography ([18F]FDG-PET) voxel intensity-based intensity-modulated radiation therapy (IMRT) for head and neck cancer. Radiother Oncol 2006;79:249e258.