Chapter 18.
ETHMOID SINUS CANCER
CASE STUDY
FILIP CLAUS, MD, PHD, WIM DUTHOY, MD, WILFRIED DE NEVE, MD, PHD
Patient History Target and Tissue Delineation
A 67-year-old male presented with nasal obstruction, Given the initial tumor extent and the difficulty of assess-
headache, and a mucopurulent postnasal drip. Rhinoscopy ing resection margins, the entire region of the ethmoid air
demonstrated mucosal congestion, and rigid nasal cells and the adjacent sinonasal cavities were felt to be at
endoscopy revealed purulent crusts. A computed tomog- risk. A compartment-related definition of the clinical tar-
raphy (CT) scan was ordered and demonstrated a poly- get volume (CTV) was used, that is, the region of the eth-
poid mass in the right ethmoidal air cells with remodeling moid sinus air cells, and the directly flanking sinonasal
of the right lamina papyracea, opacification of the eth- cavities were included. These cavities included the nasal
moidal and right frontal sinuses, and mucosal thickening cavity, the right maxillary sinus, and both sphenoidal sinus-
of the maxillary and sphenoidal sinuses. No pathological- es (the tumor invaded the posterior ethmoidal sinus cells).
ly enlarged lymph nodes were noted. The remainder of the A portion of the anterior cranial fossa was also included
workup was negative. owing to the disruption of the leptomeningeal structures.
Using a combined transfacial and neurosurgical A gross tumor volume (GTV) was not specified in this case
approach, a gross total resection was performed, and the in light of the gross total surgical resection. The CTV was
pathology was consistent with adenocarcinoma. The tumor uniformly expanded by 3 mm, generating a planning tar-
was staged as pT4cN0M0, and the patient was referred for get volume (PTV) accounting for setup uncertainty. The
postoperative radiation therapy. Owing to the close prox- CTV and PTV delineated in this patient are shown in Figure
imity of the optic pathways and lacrimal apparatus to the 18.2-1.
target volume, the patient was treated with intensity-mod- Organs at risk (OAR) in this case included the optic chi-
ulated radiation therapy (IMRT). asm, optic nerves, retinae, lacrimal glands, pituitary gland,
brainstem, brain, mandible, and both parotid glands.
Delineation of the posterior aspect of the optic nerves,
Simulation optic chiasm, and pituitary gland was based on the MRI
The patient was simulated in the supine position. A cus- dataset. The lens was not considered an OAR because
tom-made thermoplastic head cast was fabricated for cataract surgery is a minor operation with a high success
immobilization, and a knee cushion was used for patient rate and with almost no risk of complications. The optic
comfort. A planning CT scan was obtained on a Somatom pathway structures (optic chiasm, optic nerves, and both
Plus 4 scanner (Siemens Medical Systems, Munich, retinae) and brainstem were expanded to form planning
Germany). The planning CT scan extended from the ver- at-risk volumes (PRVs). The optic pathway structures were
tex of the head to the sternoclavicular junction. Over the expanded by 2 mm, whereas the brainstem was expanded
region of the paranasal sinuses, a CT slice thickness of 2 by 3 mm. Figure 18.2-1 illustrates the OAR and corre-
mm was used. Outside this region, 5 mm slices were sponding PRV in this patient.
obtained. No contrast was administered. The dataset con-
sisted of 105 transverse slices (pixel resolution 512 × 512).
A magnetic resonance image (MRI) (slice thickness 1 Treatment Planning
mm) was also obtained and registered with the planning Prior to the optimization process, three nonanatomic (vir-
CT scan to aid in the delineation of the target and nor- tual) OAR were automatically generated by subtracting from
mal tissues. the patient volume the PTV, the PTV expanded by 20 mm,
284
Ethmoid Sinus Cancer: Case Study / 285
FIGURE 18.2-1. Axial computed tomography slices illustrating the target and normal tissues delineated for this patient. Each structure (optic
nerve, optic chiasm, retina, clinical target volume [CTV]) is represented as two contours; the inner contour is the structure itself, and the outer con-
tour is the expanded structure (the planning at-risk volume, and in case of the CTV, this is the planning target volume). (To view a color version of
this image, please refer to the CD-ROM).
and the PTV expanded by 50 mm. For optimization pur- incidences were defined upfront by the treatment planner.
poses, two subvolumes of the PTV were created: a PTV with- A template of seven beams was used: five beams with the
out a buildup region (PTV-wbu), that is, the PTV without central axes located in the sagittal plane (table rotation
the subvolume within a range of 6 mm from the skin con- angle of 90 degrees and gantry angles of 0, 30, 60, 90, and
tour, and a PTV without buildup and without overlap with 330 degrees) and two beams with central axes in a trans-
OAR (PTV-wbu-woars), that is, all points inside the PTV- verse plane (table rotation angle of 0 degrees and gantry
wbu but not closer than 6 mm to the optic structures (optic angles of 75 and 285 degrees). The isocenter of all beams
chiasm, optic nerves, and retinae). Subvolumes inside the was located at the xyz-midpoint of the PTV. The energy of
PTV, that is, the PTV-wbu and the PTV-wbu-woars, were all beams was 6 MV.
used to remove conflicts during the optimization phase. The components of our planning approach include an
When the PTV and the expanded optic structures intersect, anatomy-based segmentation tool, which creates segments
a conflicting dose prescription is present. By prescribing 70 for each beam incidence.1 The segment outlines are based
Gy to the PTV subvolume outside the overlap or buildup on the beam’s eye view of the PTV and OAR. The purpose
region ( PTV-wbu-woars) and a maximal dose of 60 Gy inside of this tool is to create an increasing number of segments
this region, the conflict is removed. with decreasing distance to the optic structures and increas-
Treatment planning was performed using in-house ing thickness of overlying tissue. Sixty-seven segments were
inverse planning software. A flow diagram of the planning generated for this patient. Segment weight optimization is
process used in this patient is shown in Figure 18.2-2. Beam then performed using in-house biophysical cost function
286 / Intensity-Modulated Radiation Therapy
dosage, the International Commission on Radiation Units
CT and MR imaging
contouring and expansion and Measurements guideline of 7% was followed. The three-
dimensional maximum dose was located inside the PTV.
Table 18.2-1 summarizes the dose constraints for the
Anatomy-based segment generation OAR in this case. The dose limit for the 2 mm expanded
optic structures (optic chiasm, optic nerves, and retinae)
was 60 Gy, that is, 5% of the volume of the structure was
Segment outline
allowed to receive more than 60 Gy. Although this limit
Dose and may seem high, the normal tissue complication probabil-
computation weight optimization ity of the optic structures using our approach is low for a
number of reasons. First, the 60 Gy limit is applied to the
Not okay expanded structure. Second, the use of a biologic opti-
mization model results in an inhomogeneous irradiation
Plan evaluation
of the nerves and a low mean dose. Finally, the maximum
Okay daily fraction size is 1.7 Gy. The maximum dose for the
brainstem was a hard constraint of 60 Gy (applied to the
Treatment prescription PRV). No specific dose constraints were used for the lacrimal
data transfer to linac glands, pituitary gland, brain tissue, mandible, and parotid
glands, although biologic constraints were used for these
Dummy run and quality control structures during optimization.
treatment delivery and portal imaging Figure 18.2-3 displays the isodose distributions for the
optimized treatment plan (normalized to a median PTV
dose of 70 Gy). Dose-volume histograms of the target and
FIGURE 18.2-2. The intensity-modulated radiation therapy planning
platform consists of several modular units, including tools to generate OAR are shown in Figure 18.2-4.
segments, compute doses, optimize plans, and generate treatment pre-
scriptions. The units with gray background are in-house–developed soft-
ware programs. CT = computed tomography; MR = magnetic resonance.
Linac = linear accelerator. TABLE 18.2-1. Summary of Optimization Parameters used for
IMRT Planning
Target Volumes Physical Constraints
software.2 The segment outlines are further optimized using PTV-wbu Median dose 70 Gy, ICRU guidelines (–5% to +7%)*
a leaf position optimization scheme.3 This tool evaluates PTV-wbu-woars ICRU guidelines (–5% to +7%)
the effects of changing the position of each leaf within each Three-dimensional maximum dose has to be located inside the PTV
segment on the biophysical objective function. Leaf posi-
Organs at risk Biophysical constraints
D95 ≤ 60 Gy, biologic optimization (DVH reduction
tion changes that decrease the value of the objective func- Optic chiasm (PRV)
tion are retained. After several possible positions (eg, changes scheme†)
of 1, 3, 5, and 10 mm) have been evaluated, an external dose Optic nerve (PRV) D95 ≤ 60 Gy, biologic optimization
engine recomputes the dose distribution, based on the adapt- Retina (PRV) D95 ≤ 60 Gy, biologic optimization
ed leaf positions and weights. Only leaf position changes Brainstem (PRV) Dmax ≤ 60 Gy, biologic optimization
Brain No physical constraints, biologic optimization
that comply with the multileaf collimator (MLC) constraints Mandible No physical constraints, biologic optimization
of the linear accelerator are evaluated (ie, segments that can Virtual organs at risk No physical constraints, biologic optimization
be delivered). The remaining number of segments after
D95 = the dose that covers 95% of the organ at risk; Dmax = the maximum dose;
shape and weight optimization was 40. The segments are DVH = dose-volume histogram; ICRU = International Commission on Radiation
provided as separate MLC beams to the dose computation Units and Measurements; PRV = planning at-risk volume (ie, the expanded struc-
unit (convolution-superposition algorithm). In the last step, ture); PTV = planning target volume; PTV-wbu = planning target volume with-
the segment delivery sequence is optimized to ensure the out buildup; PTV-wbu-woars = planning target volume without buildup and without
shortest possible delivery time. For a more detailed descrip- overlap with an organ at risk.
*An underdosage of more than 5% inside the planning target volume was accept-
tion and discussion of the treatment planning process, the ed in the regions adjacent to or overlapping the optic structures (planning at-risk
reader is referred to previously published research.4,5 volumes).
The PTV prescription dose was 70 Gy, delivered in 2 Gy †The parameters for the biologic optimization were calculated according to a
daily fractions. An underdosage of more than 5% inside the dose-volume histogram reduction scheme. The volume and slope parameters to
PTV was accepted in the regions adjacent to or overlap- calculate the normal tissue complication probability were based on published
data. For the optic structures and the brainstem, the median toxic dose values
ping with the optic structures (PRVs), as well as in the buildup were decreased to customize the plan to the protocol constraints (see Claus F
region of the 6 MV photon beams. With regard to PTV over- et al5 for details).
Ethmoid Sinus Cancer: Case Study / 287
100 C3
2 mm expanded
right optic nerve PTV-wbu woars
90
PTV
80 2mm expanded
left optic nerve
70
Frequency (%)
60
2 mm expanded
optic chiasm
50
40 2 mm expanded FIGURE 18.2-3. Dose-volume histograms for
right retina the target volumes and the optic structures
30 for the presented case. A median dose of
2 mm expanded 70 Gy is prescribed to the planning target vol-
20 left retina ume (PTV). Planning constraints: C1 = maxi-
mal dose in the PTV ≤ 107% of the prescribed
10 dose; C2 = maximally 5% of the volume of the
C2 C1 optic structures is allowed to receive ≥ 60 Gy;
0 C3 = the maximal underdosage in the PTV
0 10 20 30 40 50 60 70 80
without buildup and without overlap with
Dose(Gy) an organ at risk (PTV-wbu-woars) is 5%.
FIGURE 18.2-4. Isodose distributions for the presented case in a coronal, sagittal, and transverse plane. The planning target volume (PTV) is
shown as a red solid line. A median dose of 70 Gy is prescribed to the PTV. (To view a color version of this image, please refer to the CD-ROM).
Treatment Delivery and Quality clinically implemented. For the presented case, the patient’s
treatment fields were delivered to a head phantom with an
Assurance ionization chamber located near the isocenter position. Daily
Treatment was delivered on an SLi plus linear accelerator electronic portal imaging was performed for the first five
(Elekta, Crawley, UK) equipped with the Precise MLC. This treatment fractions to reduce systematic setup errors. Weekly
MLC replaces the upper movable jaws inside the linear portal imaging was performed thereafter.
accelerator head and has 40 pairs of tungsten alloy leaves
that project a width of 1.0 cm at the isocenter. The treat-
ment delivery time was completed within a 15-minute time Clinical Outcome
slot (electronic portal imaging included). IMRT treatment was well tolerated except for the devel-
Rigorous quality assurance procedures were performed opment of grade 2 conjunctivitis (ie, symptomatic con-
to ensure accurate treatment delivery. Gel dosimetry was used junctivitis, not interfering with activities of daily living).
before the IMRT treatments for ethmoid sinus tumors were The patient also complained of fatigue. No grade 3 or 4
288 / Intensity-Modulated Radiation Therapy
toxicity was seen. At his latest follow-up (14 months post- Patient follow-up reports revealed no evidence of disease in
treatment), the patient remained without evidence of dis- 7 patients. Two patients died, one of a locoregional relapse
ease recurrence, and his only subjective ocular complaint and the other from distant metastases. No unilateral or bilat-
was occasional tearing. An ophthalmologic evaluation did eral visual impairment has been observed thus far.
not reveal any difference with the pretherapeutic findings,
that is, the vision and ocular pressure were not changed,
the fundoscopy was normal, and there were no signs of References
cataracts in either eye. 1. De Gersem W, Claus F, De Wagter C, et al. An anatomy based
Our clinical results using IMRT in patients with ethmoid segmentation tool for intensity modulated radiotherapy of head
sinus tumor were reported earlier.5 Eleven patients with and neck cancer. Int J Radiat Oncol Biol Phys 2001;51:849–59.
T1–4N0M0 ethmoid sinus cancer were treated at Ghent 2. Derycke W, De Gersem S, Colle C, et al. Inhomogeneous
target-dose distributions: a dimension more for optimization?
University between February 1999 and July 2000. Ten patients
Int J Radiat Oncol Biol Phys 1999;44:461–8.
were treated postoperatively, and one was treated definitively.
3. De Gersem W, Claus F, De Wagter C, et al. Leaf position
In 9 of 10 patients who underwent surgery, the resection was optimization for step and shoot IMRT. Int J Radiat Oncol Biol
macroscopically complete. Microscopic invasion of the resec- Phys 2001;51:1371–88.
tion margins was observed in 1 patient. Overall, treatment 4. Claus F, De Gersem W, Vanhoutte I, et al. Evaluation of a leaf
was well tolerated. Light sensitivity (grade 1) was observed position optimization tool for IMRT of head and neck cancer.
in 8 patients. Tearing, cornea, and lacrimalation symptoms Radiother Oncol 2001;61:281–6.
were observed in 10 patients (5 grade 1, 5 grade 2). Pain and 5. Claus F, De Gersem W, De Wagter C, et al. An implementation
dryness of the eyes were reported in 8 patients (6 grade 1, 2 strategy for IMRT of ethmoid sinus cancer with bilateral sparing
grade 2). None of the patients developed dry-eye syndrome. of the optic pathways. Int J Radiat Oncol Biol Phys 2001;51:318–31.
