Estimation of normal tissue complication probabilities with three-dimensional technology

Abstract

Radiation oncologists routinely make treatment decisions based, in part, on the perceived risk of normal tissue injury. Accurate assessments of the risk of normal tissue damage, however, are often elusive. Almost all of our knowledge of radiation tolerance of normal tissues was developed prior to the era of 3-dimensional (3-D) treatment planning and dose calculation and is therefore lacking in precision (5). There is a need to use the new technology to quantitatively study and better understand the physical and biological parameters related to the development of radiation-induced organ injury. The group at the University of Michigan is aggressively applying their extensive 3-D treatment planning tools to this topic. They previously reported the results of an analysis relating the risk of hepatic toxicity to the 3-D dose distribution within the liver (9). In this issue, Martel et al. quantitatively relate the 3-D radiation dose distribution within the lungs of 63 patients to the development of clinical pneumonitis (15). Their results suggest that it may be possible to calculate realistic estimates of the risk of normal tissue injury based on the 3-D dose distribution. This a very important paper and is one of the first attempts to meticulously apply 3-D technology to this important subject. Radiation-induced symptomatic pulmonary injury is common following treatment of tumors in and around the thorax and remains a major obstacle to aggressive local therapy for lung cancer (3, 4, 7, 10, 17). The study also serves to highlight many of the difficulties of applying 3-D tools to determine normal tissue complication probabilities (NTCP). Estimating the risk of organ injury requires that the radiation dose distribution be both known and interpretable. While modern radiation treatment planning tools, such as those used in this study, dramatically improve our ability to know the detailed 3D dose distribution, interpretation remains difficult. (DVH) used to display the 3-D dose information are commonly used as aids in assessing 3-D treatment plans. Dose-volume histograms may not, however, in themselves provide a clinically useful endpoint. The NTCP model used by Martel et al. is a mathematical manipulation to derive a NTCP from the DVH (11). Given the complexity of 3-D radiation dose distributions, there is great appeal in attempts to reduce this vast information into a single numerical value. This may not, however, be a realistic goal because these mathematical models ignore patient-specific biological risk factors that are likely to be important determinants of the clinical outcome. For example, the risk of developing dyspnea following thoracic irradiation must be related to the pretreatment pulmonary status. In fact, one of the patients in Martel’s study who developed grade 4 pneumonitis had a very low NTCP value of 2.2%, but had severe preexisting pulmonary disease. Therefore, clinicians should use extreme care when these models are applied in patient-care situations. Another weakness of DVHs are that they discard spatial information. It is assumed that irradiation of X% of an organ to Y Gy will result in a clinical outcome that is independent of which specific volume of the organ is irradiated. This may be true for some organs in which the function is uniformly distributed throughout the organ. It is not likely to be true for organs such as the brain, eye, or heart where the function is heterogenously distributed. This concept is well supported by radiosurgery data suggesting that the development of clinical symptoms is dependent not only on the dose-volume parameters, but on the location of the treated volume (6, 16). Even in the liver and lung where different regions of the organ normally have similar and independent functions, functional heterogeneities may be present in the diseased state. For example, our group has used single photon emission computed tomography (SPECT) lung scans to demonstrate functional heterogeneities within the lungs of patients with lung cancer or chronic obstructive pulmonary disease ( 14). These SPECT scans provide a 3-D map of functioning vascular/alveolar complexes. Pretreatment scans can be used to design radiation treatment fields that