Routine 3D Treatment Planning: Opportunities, Challenges, and Hazards

Routine 3D Treatment Planning: Opportunities, Challenges, and Hazards

ABSTRACT: Three-dimensional (3D) treatment planning refers to the use of software and hardware tools to design and implement more accurate and conformal radiation therapy. This is a major advance in oncology that should lead to the reduction of treatment-associated morbidity and facilitate safe dose escalation for many tumor sites. This technology affords the incorporation of physiologic and anatomic information into the treatment planning process, further enhancing our ability to improve the therapeutic ratio. However, as with any new technology, care must be taken when applying it in the clinic. The introduction of 3D planning presents new challenges to existing quality assurance systems. These need to be addressed to maintain patient safety. Based on our experience with over 1,500 patients treated at Duke University, the benefits, challenges, and hazards of routine 3D treatment planning are discussed. [ONCOLOGY 14(8):1191-1201, 2000]


In an ideal world, radiation treatment beams would fully encompass
the target tissue and exclude all nontarget normal tissues. In
practice, however, this is impossible due to the intimate
relationship between tumors and surrounding normal structures.
Radiation treatment planning, therefore, is based on a compromise
between tumor and normal tissue considerations. The radiation
oncologist’s traditional approach to this problem is to:

(1) Review appropriate clinical and radiographic information to
identify targets.

(2) Choose appropriate treatment beam orientations.

(3) Under fluoroscopic guidance, orient the treatment beams onto the
patient so that the fields include the defined target volume, and
obtain simulation films.

(4) Shape treatment beams based on the clinician’s knowledge of
the location of the target and normal tissues, typically by drawing
on the simulation films.

This traditional approach, however, has several limitations.

First, it is extremely difficult for most radiation oncologists to
accurately assimilate complex three-dimensional (3D) anatomic data
such as that obtained with computed tomography (CT), magnetic
resonance imaging (MRI), positron emission tomography (PET), and
single-photon emission computed tomography (SPECT), unless the target
is bound by radiographic landmarks (for example, the sella). The
inability to precisely understand the often complex 3D relationships
among a variety of structures limits the physician’s ability to
conceive of “optimal” beams.

Second, because 3D diagnostic information is usually conveyed with a
series of axial images, it is easier to conceive of beams that lie
within the transverse plane. Nonaxial beams may be therapeutically
advantageous, but they are more difficult to visualize, and may be
underutilized by the radiation oncologist.

Third, even if the “optimal” beam is conceived, it is very
difficult to accurately set up that beam in the simulator room. The
physician is limited by the need to relate the 3D diagnostic data to
the planar fluoroscopic images.

Fundamentals of 3D Treatment Planning

Three-dimensional treatment planning broadly refers to a variety of
tools and procedures that facilitate the use of 3D data during the
planning process. Different approaches to this process have been
taken.[1-4] In 1991, we implemented an approach to 3D treatment
planning at Duke University that was developed and initially
implemented at the University of North Carolina (UNC) in the 1980s.
Initial clinical results from UNC have been reported.[5-7] A brief
outline of the process used at Duke is as follows:

(1) Three-dimensional imaging (eg, CT) is obtained with the patient
in an immobilization device that is used throughout treatment. A
reference coordinate system is defined and marked on the
immobilization device (and, possibly, on the patient as well).

(2) The patient returns home and the treatment planning proceeds on
3D imaging data sets.

(3) Structures of interest, targets, and normal tissues are
identified on the images.

(4) Treatment-planning software (initially GRATIS [Sherouse Systems
Incorporated], and more recently, PLUNC [Plan University of North
Carolina]) is used to view the 3D relationship between structures of
interest from any direction (including axial and nonaxial
orientations). Information from different diagnostic imaging
modalities can be viewed.

(5) Beam orientations are selected and beams are shaped, based on the
projection of the structures of interest as seen along the
beam’s-eye view.

(6) Doses are calculated and adjustments in beam weights, wedges,
blocks, and beam orientations are made as desired in an iterative fashion.

(7) Digitally reconstructed radiographs of each beam are generated
(including the block shape and desired structures) and can be used in
lieu of physical simulator films.

(8) Setup instructions to facilitate the implementation of treatment
beams at the physical simulator and treatment machine are provided
and include field size, gantry, collimator, and table position
(relative to the predetermined coordinate system).

The 3D treatment planning tools evolved, in part, as use of the
systems identified areas that needed improvement. Recent additions to
the planning software include the use of predetermined beam
templates, viewing of real-time digitally reconstructed radiographs,
and the ability for multimodality imaging.[8]

Performing the treatment planning on the image data set reduces the
stress and associated practical limitations of having the patient on
the simulator table. Planning that considers beams from any
orientation can be performed at a comfortable pace. The information
provided by the treatment planning system parallels that provided by
the physical simulator. Thus, the physical simulation may be omitted.

In most instances, however, a physical simulation is performed to
obtain a set of simulator films (that do not have the reconstruction
artifact of the digitally reconstructed radiographs) and, in some
cases, to verify that the beams are implementable and clinically
appropriate. An additional consideration is that the physical
simulation provides an opportunity to assess
respiratory/diaphragmatic movements not readily appreciated on the CT
data set.

If the physical simulation is omitted, the first day on the treatment
machine essentially serves as the simulation and will generally take
more time than a typical treatment. Thus, the choice to skip the
session in the physical simulator is related to time restrictions on
use of the simulator vs use of the treatment machine.

The goals of this article are to:

(1) Describe our initial experience with 3D radiation treatment
planning in approximately 1,500 patients.

(2) Report the results of a physician survey assessing the perceived
utility of 3D treatment planning in 856 of these patiens.

(3) Review the rationale for, and our experience with, incorporating
physiology into the 3D planning process.

(4) Discuss some of the challenges and hazards of 3D planning.

Initial Clinical Experience and Survey Results

The number of patients who underwent 3D treatment planning per year
from 1991 through 1998 is shown in Figure
. For the 856 patients treated between March 1995 and 1998, the
treating physicians were asked to qualitatively assess the perceived
benefits of 3D treatment planning for each case. Surveys were
delivered immediately following completion of radiation therapy, and
80% of the surveys were returned.

The questions and a summary of the results are shown in Table
. The rate of affirmative answers to the questions is a minimum
value since all missing answers were assumed to be negative. The
results reported are for the entire patient population and for the
largest subsets of patients (ie, those with lung, prostate, and brain

For patients with prostate cancer, the use of 3D planning tools
altered the shape of the radiation beams, but generally not their
orientation. For patients with lung cancer, 3D planning tools led to
atypical beam orientations and gave the physician confidence to treat
patients with unconventionally high radiation doses. No notable
differences were seen over time, probably because the surveys were
started several years after the implementation of 3D planning (ie,
after the steep learning curve).

The extra time spent by the physician per prostate cancer case was
only 30 minutes, probably due to the use of a conventional
four-field box technique in almost all cases. The brain and lung
cancer cases required more time since unusual beam orientations,
typically with field reductions, were commonly used.

The utility of 3D treatment planning is not limited to dose
escalation. In fact, a better understanding of the 3D anatomic
relationships between the target and normal tissues likely improves
the treatment planning process, even when conventional beam
orientations and target doses are used. Simply placing the treatment
isocenter and beams in their intended location is a worthwhile goal,
as illustrated by the survey results: Sixty-seven percent of the
surveys indicated that the additional effort was worthwhile, even
though dose escalation was performed in 20% of cases. In about 65% of
cases, it was believed that 3D planning reduced the volume of normal
tissue irradiated and improved the therapeutic ratio. This rate was
lowest in patients with brain tumors, since they frequently were
treated with large opposed lateral fields.

Although it is beyond the scope of this article to outline the
potential benefits of 3D planning at all disease sites, our
experience with the process at certain sites deserves mention. For
cancers of the paranasal sinuses, a comparison of 3D planned beams to
conventional beam arrangements suggests a dramatic benefit
attributable to 3D planning. The software allows the use of nonaxial
beams that markedly reduce the dose to the optic structures.[9] The
3D tools facilitate the use of conformally shaped tangent fields that
limit the volume of heart irradiated in patients with breast
cancer.[10-11] We have developed beam “bouquets”—ie,
multiple noncoplanar beams that provide rapid dose gradients at
target edges similar to what can be achieved with radiosurgery

Acknowledgment: Thanks to Jane Hoppenworth for her skillful
assistance with data acquisition, and to the University of North
Carolina Department of Radiation Oncology for software support.


Loading comments...
Please Wait 20 seconds or click here to close