Twenty patients having received definitive radiation therapy for advanced H&N cancer were identified for study. Cases selected included nasopharynx, soft palate, base of tongue, tonsil, supraglottic larynx, and H&N unknown primary. All patients underwent CT imaging, with intravenous contrast, following H&N immobilization with a slice thickness of 1.25 mm. The same CT image sets were used for the generation of all plans, namely, conventional H&N radiotherapy using conformal lateral field arrangements, target defined IMRT (TD-IMRT) and conformal avoidance IMRT (CA-IMRT), respectively.
Conventional H&N Field Design
In the case of conventional H&N radiotherapy patients first underwent virtual simulation. The CT isocenter was determined and a GTV was carefully delineated. Comprehensive H&N shrinking field design (lateral large fields, off cord, and final boost), including posterior neck electron fields, was carried out using a lateral beams eye view with 2D-projection of the GTV. The regional field or CTV (appropriate nodal coverage as indicated on lateral views) was defined using anatomic boundaries dictated by the knowledge of H&N control and failure patterns established over several decades of radiotherapy (). This process is essentially identical to drawing treatment fields on lateral radiographs obtained from a conventional simulator. Organs at risk (OAR) are not specifically identified in this planning process since essentially all tissues contained within the lateral beam projections are treated.
Conventional field design for a patient with T2N2b squamous cell carcinoma of the left tonsil.
Target Defined IMRT (TD-IMRT)
In target defined IMRT (TD-IMRT) the first two steps of the virtual simulation process are identical to that of conventional H&N radiotherapy. Thereafter organs at risk (OAR) are delineated, such as optic chiasm, parotid glands, spinal cord, brainstem, larynx, mandible and anterior oral cavity. During the planning process, dose objectives are placed on each OAR in an effort to diminish overall normal tissue complications.
The next step in TD-IMRT involves the definition of high risk and low risk clinical target volumes, CTV1 and CTV2, respectively. CTV1 and CTV2 are generated by identification and careful delineation of high risk and low risk nodal regions on each consecutive CT image. Once the GTV, CTV1, CTV2, and OARs are defined, appropriate planning target volumes (PTV) for each target are created by volumetric expansion. If a particular PTV overlaps an OAR, a “residual” OAR is created by subtracting the overlap between OAR and PTV. Thereafter, dose objectives are selected for all PTV and OAR (or residual OAR) targets and then the inverse planning process is initiated.
Conformal Avoidance IMRT (CA-IMRT)
In conformal avoidance IMRT (CA-IMRT) the first three steps of the virtual simulation process are identical to that of TD-IMRT. First, one identifies the CT isocenter and then delineates the GTV and OARs, which we will also refer to as conformal avoidance structures. However, in contrast to TD-IMRT, CTV1 and CTV2 are not defined by outlining individual nodal levels on consecutive CT slices for CA-IMRT. Rather, these elective targets are initially defined by simply projecting the conventional lateral treatment fields appropriate for the specific anatomical H&N tumor and stage. This represents the critical distinction between the two IMRT design methods. Rather than slice-by-slice contouring of three-dimensional nodal stations from CT anatomy, the physician simply uses his/her own conventional lateral fields as the guiding template for elective CTV design and then subtracts from this volume those normal tissue OARs where no tumor risk is present (e.g. spinal cord, parotid gland). We now describe in detail the definition of CTV1 and CTV2 using conventional lateral Head and Neck treatment field designs.
Generation of CTV1 and CTV2 using conventional lateral H&N treatment fields
This method proceeds along the following five steps. Step 1: based on several decades of collective clinical experience (13
) the conventional lateral fields appropriate for the specific H&N cancer patient are designed. In step 2, the GTV and key conformal avoidance structures such as parotid glands and spinal cord are contoured. In step 3, a crude maximal treatment volume is generated that reflects the full lateral H&N beam coverage by selecting the 85% isodose line as a hypothetical prescription line. Step 4 consists of generating a region of interest that represents the full volume of irradiation as specified by the isodose volume selected in step 3. In step 5, the final CTV is generated by subtracting avoidance structures from the region of interest, selective trimming of areas deemed not to contain lymphatic risk for tumor spread, and then dividing the resulting CTV into high risk CTV1
and low risk CTV2
as applicable. In the following section, we describe each of these five steps in greater detail.
The first step in the CA-IMRT design process is the most familiar to practitioners of H&N radiation oncology, namely conventional lateral field design. To illustrate the CA-IMRT method, we highlight a patient with a T2 N2b M0 squamous cell carcinoma of the left tonsil. depicts the conventional lateral field design created for this patient.
In step 2, the GTV and key conformal avoidance structures are contoured. The conformal avoidance structures (OARs in the case of TD-IMRT) such as the parotid glands, eyes, optic chiasm, larynx, mandible, posterior fossa, anterior oral cavity are delineated on each CT slice just as with TD-IMRT. The spinal cord is drawn slightly differently, since, we aim to conformally avoid not only the spinal cord but also the spinal column, which harbors minimal risk for H&N cancer spread, we have decided to approximate the spinal column by an inverted U-shape that includes the spinal cord, the vertebral body and paraspinal muscles. One first draws this inverted U-shape on 7 to 9 representative transverse CT-slices through out the volume, and the spinal column conformal avoidance region is then generated through interpolating between these 7 to 9 inverted U-shaped contours (cf. ). Skin (defined as 4 mm thick ring) is also defined as an avoidance structure. shows several of these conformal avoidance structures as defined.
Contouring of the GTV and the conformal avoidance structures for a patient with squamous cell carcinoma of the left tonsil. OAR shown are the parotid glands (P), mandible, and spinal column.
In order to initiate definition of an elective nodal treatment volume, we then generate a pre-optimized treatment volume by employing the conventional lateral fields designed in step 1 of this process. By selecting the 85% isodose line, we generously approximate the prescription volume commonly chosen for H&N cancer patients treated with conventional lateral beam techniques (90–100% common at most institutions).
Following selection of the 85% isodose line, the preliminary volume for CA-IMRT is generated as follows. The upper and lower contours of the elective nodal volume are coincident with the superior and inferior field borders as indicated by the 85% isodose line on the sagittal view. One can approximate the shape of 85% isodose line in the sagittal view quickly by drawing additional contours in locations where the gradient of the 85% isodose line changes in the sagittal view. Interpolating between these contours then generates the entire elective nodal volume (cf. ). However, some treatment planning systems allow one to directly convert an isodose volume into contours and this feature can then be used, instead of above described contouring procedure, to generate the preliminary volume for CA-IMRT.
The contouring of the elective nodal volume as indicated by 85% isodose volume is shown.
Next one subtracts all avoidance structures from the elective nodal volume. This yields the raw elective nodal volume shown in . One then proceeds to generate the residual total elective nodal volume shown in by first trimming away the part of the raw elective nodal volume that lies on the outside of the mandible next the fingers around the avoidance structures that are generated when these structures are subtracted from the raw elective nodal volume are trimmed away. This residual total elective nodal volume should not simply be accepted as is, but should be carefully reviewed slice by slice and if necessary adjusted and modified according to disease site and nodal involvement, and such that it contains the entire GTV.
Once one has arrived at a satisfactory residual total elective nodal volume, high risk and low risk clinical target volumes are designed as follows. First one divides the residual elective nodal volume into two parts: a high risk and low risk nodal volume (cf. ) using the appropriate contouring tool. Doing the following then generates a separate high-risk CTV1 and low risk CTV2. One first generates a copy of the divided residual clinical target volume. The high-risk CTV1, is then obtained by deleting all contours that are part of the low risk nodal volume form one of the copies of the divided residual elective nodal volume. The low risk CTV2 is obtained by subtracting the high risk CTV1 from the remaining intact copy of the divided residual elective nodal volume. The remaining intact copy of the divided residual elective nodal volume can then be deleted, since it has been split into a high risk CTV1 and low risk CTV2.