Chapter 19.

TARGET DEFINITION IN
NON–SMALL CELL LUNG
CANCER CASE STUDY
THOMAS GUERRERO, MD, PHD, YERKO BORGHERO, MD, CRAIG W. STEVENS, MD, PHD

Patient History ter was set at the time of the planning. All CT data sets were
acquired on a commercial multislice helical CT scanner
A 61-year-old male smoker developed a persistent respi-
(MX8000 IDT, Philips Medical Systems, Andover, MA).
ratory infection 1 month prior to presentation. A chest
An external fiducial marker was placed on the patient’s
radiograph was obtained and revealed a right middle lobe
abdomen, and a video tracking system monitored the res-
lung opacity suspicious for malignancy (Figure 19.1-1). A
piratory phase and relative respiratory effort of the patient
computed tomography (CT) scan demonstrated a 3.2 cm
during the simulation session (Real-time Position
mass in the right lower lobe and a 1 cm nodule in the right
Management system, version 1.5.1, Varian Medical Systems,
hilum. A positron emission tomography (PET) study
Palo Alto, CA). The patient was provided with video feed-
revealed hypermetabolic activity in the right middle lobe
back of the respiratory cycle and was coached on normal
mass. No suspicious activity was seen in the hilum, in the
breathing to provide a regular breathing pattern. The patient
mediastinum, or outside the thorax. A CT-guided fine-
was also coached on breath-hold techniques to maintain
needle aspiration was performed, and the pathology was
a normal inspiration and expiration breath-hold for 10 to
consistent with poorly differentiated non–small cell car-
15 seconds (required for a fast thoracic CT acquisition).
cinoma.
The patient setup is illustrated in Figure 19.1-2A, and a
Owing to his poor pulmonary function, the patient was
sample of his regular breathing pattern is shown in Figure
felt to be ineligible for surgical resection and was referred
19.1-2B.
for definitive radiation therapy (RT). The remainder of his
Three CT scans, with 3 mm slices, were obtained for
metastatic workup was negative. He was thus staged with
this portion of the treatment planning session to define
T2N0M0 (stage IB) disease.

Simulation
The patient was immobilized in the supine position. The
breath-hold CT imaging technique for determination of
the internal margin owing to respiratory motion was used
and is described here.1,2 The patient was immobilized using
a Vac-Lok device (MED-TEC, Orange City, IA) with his
arms above his head grasping a T-bar, designed to reduce
setup uncertainty. The patient was aligned with the axis of
the scanner based on a CT scout film, and three radiopaque
markers were placed at the level of the carina for identifi-
cation of the isocenter reference point. The treatment isocen-

Editor’s note: This case was not treated with intensity-modulated

radiation therapy. It is presented here to illustrate important issues FIGURE 19.1-1. Posterior-anterior (A) and lateral (B) chest radiographs
surrounding target delineation in patients with non–small cell obtained prior to treatment. Arrows indicate the location of the 3.2 cm
lung cancer undergoing radiation therapy. mass in the lateral radiograph.

358
 Target Definition in Non–Small Cell Lung Cancer: Case Study / 359

TABLE 19.1-1. Comparison of CT Parameters Used for Free-

Breathing Slow CT versus Breath-Hold Fast CT Scans Used in
the Planning of Thoracic Radiation Therapy
1m
CT Detector Rotational Table Acquisition
Technique Pitch Configuration Speed Speed Time

Slow CT 0.33 8 × 3 mm 1.0 s/rotation 0.80 cm/s 125 s

Fast CT 1.5 8 × 3 mm 0.42 s/rotation 8.6 cm/s 11.6 s

CT = computed tomography.

inspiration BH-CT (iBH-CT) and expiration BH-CT (eBH-

CT) images were obtained at normal breathing efforts (table
pitch 1.5). The monitored respiratory traces obtained are
shown in Figure 19.1-2.
The CT planning session was coordinated with the
patient’s staging PET-CT imaging session so that the immo-
bilization devices would be available for data acquisition.
The PET-CT imaging session was performed on a GE
Discovery ST PET/CT scanner with a 74 cm bore (GE
Healthcare, Milwaukee, WI), which accommodates the use
of thoracic immobilization devices. The PET images were
acquired with 3 minutes per couch position, for a total of
five couch positions or 87.3 cm axial length, with the patient
undergoing normal, quiet breathing. The CT component
(PET/CT-CT) was obtained with the patient instructed
to hold his breath at midinspiration, although no feedback
guidance was used. The registration of the PET emission
and the CT image sets was verified at the time of the aqui-
sition because the CT image set was also used for the atten-
uation correction of the PET emission data prior to
reconstruction.
FIGURE 19.1-2. Respiratory feedback. The patient was positioned supine The entire set of imaging studies was sent to the plan-
on a Vac-Lok bag and wing board with his arms above his head, where ning workstation running Pinnacle3, version 6.2b (Phillips
he grasps a T-bar (A). The video monitor used for feedback guidance and Medical Systems, Andover, MA). A region of interest defin-
the three respiratory traces are shown. Free-breathing (FB) computed ing the spine and vertebrae was defined on the FB-CT, as
tomography (CT) has a prolonged expiratory phase indicative of this shown in Figure 19.1-3A and Figure 19.1-3B. A CT-to-
patient’s chronic obstructive pulmonary disease (B). The inspiration CT 6 degrees of freedom rigid body registration was per-
breath-hold (BH-insp) CT (C) and the expiration breath-hold (BH-exp) CT formed between FB-CT and the eBH-CT, iBH-CT, and
(D) were obtained at the time indicated by the horizontal bar. (To view
PET/CT-CT using a mutual information algorithm.3 The
a color version of this image, please refer to the CD-ROM.)
FB-CT was the primary set in each case. A resulting regis-
tered image pair is shown in Figure 19.1-3C and 19.1-3D.
The result of the registration of the iBH-CT and eBH-CT
the internal target volume (ITV). Two CT acquisition modes with the FB-CT is illustrated in Figure 19.1-4.
(fast and slow) were used, and their parameters are given
in Table 19.1-1. The fast and slow acquisition modes scan
1 meter in 11.6 and 125 seconds, respectively. The first Target and Tissue Delineation
CT image acquisition was obtained with the patient breath- The breath-hold CT-based target delineation process used
ing normally or freely breathing (FB-CT) using slow acqui- at our institution is described first.1,2 The gross tumor vol-
sition parameters (table pitch 0.33). The scan extended ume (GTV) was defined as the volume of radiographical-
from the mandible to the tip of the liver to include the entire ly apparent disease, including both the primary tumor and
thoracic cavity and upper abdomen. The patient was next involved nodes. The GTV was outlined on transaxial images
instructed on breath-holding at normal inspiration and at on the FB-CT, eBH-CT, and iBH-CT image volumes. In
normal expiration for the subsequent breath-hold (BH- cases such as this, in which the tumor is completely sur-
CT) CT acquisitions. The fast CT acquisition was used, and rounded by normal lung, the initial contour set was drawn
360 / Intensity-Modulated Radiation Therapy

FIGURE 19.1-3. Image registration. This

process assumes a stationary spine and uses
mutual information to perform a 6 degrees of
freedom registration. In (A) a simple square
contour surrounds the vertebral body on this
transaxial image from the free-breathing com-
puted tomography (FB-CT) volume, and (B)
shows a sagittal view through the entire image
and contour set. The sagittal images from the
FB-CT (C) and the expiration breath-hold com-
puted tomography (eBH-CT) (D) are registered.
The eBH-CT is resliced to correspond to the
registration results with the FB-CT as a ref-
erence. (To view a color version of this image,
please refer to the CD-ROM.)

FIGURE 19.1-4. The registered free-breath-

ing computed tomography (FB-CT), expiration
breath-hold computed tomography (eBH-CT),
and inspiration breath-hold computed tomog-
raphy (iBH-CT) are shown in coronal (A–C) and
sagittal (D–F) cross-sectional images. The FB-
CT images (A and D) show the tumor (arrow)
smeared over the motion range limits indicat-
ed by the dotted line. The tumor and normal
structures demonstrate motion artifact, which
appears as the nonuniform modulation. The
eBH-CT images (B and E) show the tumor
(arrow) adjacent to the upper dotted line. The
iBH-CT images (C and F) show the tumor (arrow)
at the lower dotted line.

using lung window and level values on the CT image dis- green contour shown in sagittal section in Figure 19.1-5
play. In tumors contiguous with the hilum, mediastinum, represents the eBH-CT–determined contour (see Figure
or chest wall, two passes should be made in delineating the 19.1-5A), and the red contour represents the iBH-CT–deter-
contours, the first with the lung-optimized display and the mined contour (see Figure 19.1-5B). These two contours
second with mediastinal soft tissue–optimized display. The are shown to completely envelop the tumor present on the
 Target Definition in Non–Small Cell Lung Cancer: Case Study / 361

FIGURE 19.1-5. Computed tomography and positron emission tomography (PET) registration. The coronal sections illustrating the gross tumor vol-
ume (GTV) from the expiration breath-hold computed tomography (eBH-CT) (green) and inspiration breath-hold computed tomography (iBH-CT) (red)
scans. (A) A coronal section through the eBH-CT (note that the green contour encloses the tumor). (B) A coronal section through the iBH-CT, with the
red contour enclosing the tumor. (C) A coronal section through the free-breathing computed tomography; note that the tumor extends into both con-
tours, although the extent of the combined breath-hold–derived GTV is greater. (D) A registered coronal section through the fluorodeoxyglucose (FDG)-
PET scan, now that the FDG avid tumor is nearly completely enclosed by the breath-hold–derived GTV. (To view a color version of this image, please
refer to the CD-ROM.)

slow CT (see Figure 19.1-5C) and on the fluorodeoxyglu- non–small cell lung cancer (NSCLC) pathologic specimens.4
cose PET (see Figure 19.1-5D). The margin necessary to include 95% of the microscopic
The clinical target volume (CTV) was defined as the GTV extension has been reported as 6 mm for squamous carci-
plus margin to account for subclinical or microscopic dis- noma and 8 mm for adenocarcinoma. Typically, we have
ease extension. Giraud and colleagues provide an excellent used an 8 mm expansion as the default (as in this poorly dif-
analysis of the microscopic extension observed in 70 ferentiated NSCLC). The expansion of the GTV to the CTV
362 / Intensity-Modulated Radiation Therapy

should include knowledge of anatomic spread patterns.5 For shape, and position owing to respiratory motion can then
example, the microscopic spread will not occur freely across be included in the internal margin. Three methods have been
pleural or other anatomic boundaries unless there is inva- reported to measure the resulting ITV: gated or breath-hold
sion of adjacent structures. In this case (Figure 19.1-6), the CT imaging,1,2 slow CT imaging,6 and PET imaging.7 Using
tumor was positioned against the thoracic vertebral body the BH-CT imaging approach, the ITV should be formed as
and rib; the expansion should not cross the lung pleura. The the combination of the nonuniform expanded CTVs from
uniform three-dimensional expansion of the GTV drawn the iBH-CT, eBH-CT, and FB-CT. The iBH-CT and eBH-
on the iBH-CT and the eBH-CT (see Figure 19.1-6B) required CT account for the extremes of breathing, and the FB-CT
editing to remove their extension across the pleural anatom- accounts for lateral extent of motion in midcycle. Finally, the
ic boundary (see Figure 19.1-6C). The resulting nonuniform expansion to the planning target volume (PTV), which
expansion (see Figure 19.1-6D and Figure 19.1-6E) may then includes the effects of setup uncertainty and interfraction
be combined to form the ITV (see Figure 19.1-6E). target positioning error, proceeds from the ITV. The posi-
The internal margin has been designed to account for tioning uncertainties were measured at our institution8;
variations in size, shape, and position of the CTV in relation hence, the PTV is derived from the ITV plus 1.0 cm (see
to anatomic reference points. The variation of the CTV size, Figure 19.1-6F).
The second internal margin method can simply use
expansions from the GTV drawn on the slow FB-CT study
applied in two steps: first the expansion to the CTV and
then the expansion to the PTV. As before, the expansion to
the CTV should follow anatomic spread patterns. The over-
lap into the bony spine and ribs should be edited from a
uniform expansion (Figure 19.1-8). However, because of
motion within the FB-CT image set, any expansion into
the mediastinum or anterior chest wall should remain.

FIGURE 19.1-6. Clinical target volume (CTV) and planning target vol-
ume (PTV) expansion using breath-hold computed tomography (CT). In
(A) the gross tumor volume (GTV) contours derived from expiration FIGURE 19.1-7. Clinical target volume (CTV) and planning target vol-
breath-hold CT (eBH-CT) and inspiration breath-hold CT are shown super- ume (PTV) expansion using free-breathing computed tomography (FB-
imposed on a coronal section through the free-breathing computed tomog- CT). (A) The gross tumor volume (GTV) contours are delineated based on
raphy. Uniform expansion of the contours of 8 mm results in the contours the FB-CT. Uniform expansion of the contours by 8 mm results in the con-
seen in panels (B) and (C) (note the overlap into the vertebral body and tours overlapping the vertebral body and adjacent rib. (B) Manual edit-
adjacent rib). In (D) the areas of the expanded GTV that extend across ing of the expanded GTV contours results in the nonuniform expanded
anatomic boundaries were edited. (E) is a coronal section through the CTV (outer contour). (C) A uniform expansion of 1.5 cm to account for
two nonuniformly expanded CTVs obtained from the eBH-CT and iBH- internal margin and setup error results in the PTV. (D) The GTV and PTV
CT. The resulting PTV is shown in (F) . (To view a color version of this are shown in coronal section. (To view a color version of this image,
image, please refer to the CD-ROM.) please refer to the CD-ROM.)
 Target Definition in Non–Small Cell Lung Cancer: Case Study / 363

These structures may contain the lung lobe for a portion to explicitly include motion. The PTV would again be con-
of the respiratory cycle. The posterior chest wall and spine structed using a 1.0 cm margin to account for setup uncer-
do not move. An uncertainty of 0.5 cm can then be added tainty. The resulting three PTVs are compared in Figure
to the CTV to account for residual tumor motion not seen 19.1-9. We are investigating the significance of the differ-
on the FB-CT. The PTV would be constructed by adding ences between these three methods to define the internal
1.0 cm to the ITV (see Figure 19.1-8D). margin and subsequent PTV.
The third internal margin method assumes that the range
of motion would be captured in the prolonged PET acqui-
sition, in which the emission acquisition may range from Treatment Planning and Treatment
3 to 5 minutes per couch position. The emission acquisi- Delivery Issues
tion would be ungated and with the patient freely breath-
This case was planned and treated using three-dimensional
ing. The GTV for this method would then be drawn on the
conformal RT owing to the large amount of tumor motion
FB-CT and coregistered PET image sets (Figure 19.1-9A).
observed during the treatment planning session and the
Object size depends on the choice of window and level
issue of dose uncertainty owing to the interplay between
for the PET emission images and results from low spatial
tumor motion and IMRT delivery.10–12 This patient was
resolution of the PET emission images. This effect on object
treated to 66 Gy in 33 fractions.
size and activity quantification was described earlier.9 The
resulting contour set would then be assumed to include the
internal margin, and no additional expansion is required Clinical Outcome
At 4 months following completion of treatment, the patient
is alive and well, with no evidence of disease progression,
recurrence, or distant metastasis. A follow-up CT imaging
study (Figure 19.1-10) revealed that the tumor had a par-
tial response, with a greater than 50% reduction in tumor
diameter.

References
1. Forster KM, Stevens CW, Liao Z, et al. Defining the internal
target volume (ITV) using respiratory-gated CT image data
sets. Int J Radiat Oncol Biol Phys. 2004. [In press]
2. Shih HA, Jiang SB, Aljarrah KM, et al. Planning target volume
determined with fused CT images of fast, breath-hold, and
four second simulation CT scans to account for respiratory
movement in 3D-CRT in lung cancer. Int J Radiat Oncol
Biol Phys. 2004;60:613–22.
3. Studholme C., Hawkes DJ, Hill DLG. A normalised entropy
measure for multi-modality image alignment. Proc SPIE
Med Imaging 1998;3338:132–42.
4. Giraud P, Antoine M, Larrouy A, et al. Evaluation of
microscopic tumor extension in non-small-cell lung cancer
for three-dimensional conformal radiotherapy planning. Int
J Radiat Oncol Biol Phys 2000;48:1015–24.
5. Armstrong JG. Target volume definition for three-dimensional
conformal radiation therapy of lung cancer. Br J Radiol
1998;71:587–94.
6. Lagerwaard FJ, Van Sornsen de Koste JR, Nijssen-Visser MR,
et al. Multiple “slow” CT scans for incorporating lung tumor
FIGURE 19.1-8. Clinical target volume (CTV) and planning target vol- mobility in radiotherapy planning. Int J Radiat Oncol Biol
ume (PTV) expansion using positron emission tomography (PET)-derived Phys 2001;51:932–7.
gross tumor volume (GTV). (A) and (B) The GTV is delineated based on 7. Caldwell CB, Mah K, Skinner M, et al. Can PET provide the 3D
the fused PET data as shown. (C) Uniform expansion of the contours by extent of tumor motion for individualized internal target
8 mm results in the contours overlapping the vertebral body and adja- volumes? A phantom study of the limitations of CT and the
cent rib (arrows). (D) Manual editing of the expanded GTV contours results promise of PET. Int J Radiat Oncol Biol Phys 2003;55:1381–93.
in the nonuniform expanded CTV (outer contour). (E) and (F) The result- 8. Forster KM, Stevens CW, Kitamura K, et al. Changes of
ing PTV created and original GTV are shown. (To view a color version tumor motion patterns during a course of radiation therapy
of this image, please refer to the CD-ROM.) for lung cancer. Int J Radiat Oncol Biol Phys 2003;57:1–7.
364 / Intensity-Modulated Radiation Therapy

FIGURE 19.1-9. Comparison of the resulting planning target volumes (PTVs) obtained from the three methods described to account for internal motion:
the breath-hold computed tomography (CT) (red), the slow CT (green), and the PET imaging (blue). In this example, the PTVs are quite similar, irre-
spective of the method used to generate the PTV. This is usually not the case. (To view a color version of this image, please refer to the CD-ROM.)

FIGURE 19.1-10. Pre- and post-treatment computed tomography (CT)

images. Pre- and post-treatment transaxial CT images with the maxi-
mum tumor diameter (A) prior to treatment (43.9 mm) and (B) 1 month
after treatment (21.2 mm). The tumor has responded modestly to treat-
ment.
 Target Definition in Non–Small Cell Lung Cancer: Case Study / 365

9. Hoffman EJ, Huang SC, Phelps ME. Quantitation in positron 11. Bortfeld T, Jiang SB, Rietzel E. Effects of motion on the total
emission computed tomography: 1. Effect of object size. J dose distribution. Semin Radiat Oncol 2004;14:41–51.
Comput Assist Tomogr 1979;3:299–308. 12. Bortfeld T, Jokivarsi K, Goitein M, et al. Effects of intra-fraction
10. Jiang SB, Pope C, Al Jarrah KM, et al. An experimental motion on IMRT dose delivery: statistical analysis and
investigation on intra-fractional organ motion effects in simulation. Phys Med Biol 2002;47:2203–20.
lung IMRT treatments. Phys Med Biol 2003;48:1773–84.