Normal Tissue Tolerance Dose Metrics for Radiation Therapy of Major Organs Michael T. Milano, MD, PhD, Louis S. Constine, MD, and Paul Okunieff, MD Late organ toxicity from therapeutic radiation is a function of many confounding variables. The total dose delivered to the organ and the volumes of organ exposed to a given dose of radiation are 2 important variables that can be used to predict the risk of late toxicity. Three-dimensional radiation planning enables accurate calculation of the volume of tissue exposed to a given dose of radiation, graphically depicted as a dose-volume histogram. Dose metrics obtained from this 3-dimensional dataset can be used as a quantitative measure to predict late toxicity. This review summarizes the published clinical data on the risk of late toxicity as a function of quantitative dose metrics and attempts to offer suggested dose constraints for radiation treatment planning. Semin Radiat Oncol 17:131-140 �� 2007 Elsevier Inc. All rights reserved. Teffort he clinical practice of radiation oncology is a directed to effectively irradiate overt tumor and suspected sites of subclincal disease while sparing normal tissues. An understanding of normal tissue radiation tolerance is imper- ative for the determination of the aggressiveness of therapy and for the potential to ameliorate or prevent damage. A classic text published in 1968 by Rubin and Cassarett1 provided a foundation for the field of late effects in radiother- apy. This treatise consolidated existing data on radiation tol- erance of normal tissues and proposed the concept of sub- clinical injury. In 1991, Emami and coworkers2 published a landmark article in which the tolerance dose of normal tis- sues was exhaustively reviewed. In that article the 5% risk of complication within 5 years (TD5/5) and 50% risk of com- plication within 5 years (TD50/5) were reported as a function of volume of normal tissue irradiated (1/3, 2/3, or entire volume). These tolerance doses were based on a comprehen- sive literature review as well as expert opinion. Another com- prehensive historical review is available in the published pro- ceedings of the 1992 Late Effects of Normal Tissues (LENT) Consensus Conference.3 Within the past 2 decades, 3-dimensional planning has become standard, allowing the clinician to accurately quan- titate the volume of tissue receiving a given dose, graphically represented as a dose-volume histogram (DVH). Intensity- modulated radiation therapy (IMRT) with inverse planning allows the clinician the ability to contour higher-dose isovol- umes so as to avoid normal structures at the expense of greater monitor unit delivery and thus greater delivery of low-dose radiation. As a result, the concept of normal tissue tolerance is changing, with the volume of tissue receiving greater than a given dose (or several dose points) being a critical variable(s). An understanding of the risk of late tox- icity based on the dose-volume parameters is now becoming more clinically relevant with the advent of active investiga- tion of dose escalation in many disease sites. Certainly, there is a spectrum of confounding variables that can impact normal tissue tolerance. Broadly, these can be classified as treatment-, host-, organ-, or tumor-related vari- ables. Examples of treatment-related variables include total and fractional dose, dose rate, overall treatment time, treat- ment energy, treatment volume, the use of concurrent che- motherapy, radiation protectors or other biological modifi- ers, and the interval between radiation courses in patients undergoing a second course of radiation. Host-related vari- ables include comorbid conditions (eg, diabetes and collagen vascular disease), host response to radiation, and patient age. Organ-related variables include preradiation organ compro- mise or loss, development of severe acute toxicity (resulting in consequential late effects), regional variation of radiosen- sitivity within an organ, and hierarchal organization of the organ (ie, whether damage to a portion of the organ affects only that portion or has more widespread effect). Further- more, an organ may have more than 1 type of late toxicity that may or may not have different tolerance doses. Tumors can Department of Radiation Oncology and James P. Wilmot Cancer Center, University of Rochester School of Medicine and Dentistry, Rochester, NY. Address reprint requests to Michael T. Milano, MD, PhD, Department of Radi- ation Oncology and James P. Wilmot Cancer Center, University of Rochester School of Medicine and Dentistry, 601 Elmwood Avenue, Box 647, Roch- ester, NY 14642. E-mail: Michael_Milano@urmc.rochester.edu 131 1053-4296/07/$-see front matter �� 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.semradonc.2006.11.009

infiltrate into normal tissues, either at presentation or after treatment (ie, local failure), compromising organ function and leading to late sequelae. Indeed, the concept of normal tissue tolerance is complicated, and therefore educated clin- ical judgment should supersede published constraints and expert opinion. This review will focus on recently published data relevant to late toxicity in the adult population treated with fraction- ated radiation by examining studies using 3-dimensional dose-volume metrics, including mean organ dose and DVH points. Thus, our discussion will be limited to those organ sites in which these data are available. The term VX reflects the volume of tissue (generally a percentage) receiving greater than X Gy. The doses discussed are assumed to be in standard 1.8 to 2 Gy fractions unless otherwise stated. Sev- eral toxicity scales are commonly used, with subtle differ- ences between them. For this review, we favor the RTOG/ EORTC late toxicity criteria3 because the majority of available data are reported by using these criteria. Table 1 summarizes the Radiation Therapy Oncology Group/European Organiza- tion for Research and Treatment of Cancer (RTOG/EORTC) late toxicity scale for those organ sites discussed later. In the Common Terminology Criteria for Adverse Events,4 no at- tempt is made to attribute toxicity to an individual treatment modality (radiation, surgery or chemotherapy) because this distinction may be difficult in the era of concurrent combined modality treatment. Likewise, the distinctions between early and late effects of treatment are less obvious using the Com- mon Terminology Criteria for Adverse Events because the interval between treatment and onset of complications is not considered. Table 2 summarizes the dose metrics discussed in the text later. Figure 1 depicts the risk of late organ injury as a func- tion of dose and volume. Spinal Cord Classically, the spinal cord TD5 is often reported to be around 45 to 50 Gy, although it is generally accepted to be higher. Nevertheless, radiation oncologists often limit spinal cord maximal doses to 45 to 50 Gy because a 5% risk of grade 2 spinal cord toxicity is unacceptable (with the possible exception of tumor progression leading to inevitable cord compression). There is little published on dose-volume con- straints of spinal cord because the resultant toxicity risk as- sociated with escalating doses to the spinal cord is unaccept- able. Certainly, a small volume of spinal cord exceeding maximal cord tolerance (which could result from IMRT plan- ning) may result in no discernable late toxicity, but there are no experimental data in humans to confirm this hypothesis. There are data suggesting that the cervical spinal cord has a greater radiation tolerance than the thoracic or lumbar spinal cord.5 Massachusetts General Hospital investigators recently studied 85 patients treated to the cervical spinal cord with doses in the range of 45 to 59.4 Gy (1.5 Gy equivalent frac- tions) equivalent uniform dose, 42 to 57.5 Gy maximal dose to cord center, and 57 to 74 Gy maximal dose to cord sur- face.6 Fifteen percent of these patients experienced grade 1 to 2 late spinal cord toxicity, and 5% experienced grade 3 tox- icity. Toxicity was not significantly correlated with cord length, cord volume, maximal dose to cord center, maximal dose to cord surface, or effective uniform dose. The authors conclude that an equivalent uniform dose to the cervical cord of 60 Gy in 1.5 Gy fractions or 52.5 Gy in 2 Gy fractions is safe. Brain Data are lacking on dosimetric parameters correlating with late toxicity in adult patients, although data are emerging on survivors of childhood brain tumors treated with radiother- apy.7 Recent prospective studies in adults have shown that, with partial (and limited) brain irradiation in the dose range of 50 to 60 Gy, there is minimal to no discernable effect on memory and cognition.8-12 Late radiation effects from stereo- tactic radiosurgery are a function of target volume, dose, and conformality index (which is reviewed elsewhere).13-18 Brainstem A series from Massachusetts General Hospital showed an in- creased risk of late toxicity associated with the maximal de- livered dose, V50, V55, V60, history of diabetes, and 2 surgical procedures of the base of skull on univariate analy- sis.19,20 On multivariate analysis, only V60, history of diabe- tes and 2 surgical procedures remained significant. A V60 0.9 mL versus 0.9 mL resulted in toxicity-free survival of 96% versus 79% (P .0001) and, on multivariate analysis, resulted in an 11.4 risk ratio (P .001). Parotid Parotid glands generate 60% of saliva and the majority of serous saliva, with the remainder of saliva secreted by sub- mandibular, sublingual, and minor salivary glands. Because of the proximity of the parotid to the level 2 lymphatics, IMRT is now being used in the treatment of head and neck patients to reduce the parotid dose in an attempt to prevent xerostomia. The University of Michigan group serially mea- sured, for up to 12 months, stimulated and unstimulated salivary flow directly from individual parotid glands and showed significantly better parotid functional preservation with a mean parotid dose below 24 Gy (for unstimulated flow) to 26 Gy (for stimulated flow).21-23 Mean doses corre- lated with V15 of 67%, V30 of 45%, and V45 of 24%. The TD50 was 28.4 Gy. Washington University researchers also examined whole salivary flow their technique differed from that from Univer- sity of Michigan in that they measured flow from all glands, although they included only patients whose submandibular glands received 50 Gy in an attempt to minimize this con- founding variable.24,25 They found an exponential reduction in salivary flow of 0.04/Gy of mean parotid dose (ie, e[ 0.04* mean dose]), in contrast to Michigan, in which a dose- response was not seen. This equates to a posttreatment value of 50% of pretreatment salivary flow with a mean parotid 132 M.T. Milano, L.S. Constine, and P. Okunieff