Spinal cord radiation “tolerance”: Doctrine versus data

Abstract

Listen

Translate article

Currently accepted practice regarding radiation doses delivered incidentally to the spinal cord is evidence of more radiation oncologists’ intolerance of radiation myelopathy than of the spinal cord’s intolerance to radiation. Furthermore, an excessive dependence upon published or personal anecdotes has generated a mythology that has moved the discussion of spinal cord response away from scientific dialogue and toward catechism. However, in suggesting “that for fractions of less than 200 cGy given once a day or 120 cGy given twice a day, and a total dose of 5500 cGy or less, the risk of permanent neurologic damage is very low” (emphasis added), Marcus and Million’3 have returned to the idea of making conclusions based on data rather than beliefs. In this editorial certain dogma will be critically examined in light of supporting and opposing evidence. In radiotherapy, the term “tolerance” is variously used to describe a safe dose, an acceptable dose, a dose yielding 5%, 50% or some other frequency of response. That there is no uniformly accepted definition reflects differences in the clinical “acceptability” of different types of treatmentrelated mordidity. On the other hand, in statistics, the term “tolerance” has an unequivocal definition: the tolerance dose is that above which an individual will exhibit a response* and does not apply to the entire population at risk. The dose response function (the probability of injury as a function of dose) reflects the distribution of tolerance doses over the population at risk. A single tolerance dose for a population is meaningful only when the incidence of injury is 0% below this dose and goes to 100% at the tolerance dose, that is, the dose response is a step function. Thus, to say a dose to the spinal cord of 45 Gy in 23-25 fractions represents cord tolerance is true only insofar as most radiotherapists accept its use and very few will tolerate (in practice) a higher dose. There are many published reports that indicate how far 45 Gy at 1.8-2.0 Gy per fraction is from the dosesensitive region of the dose-response curve. The accompanying table lists incidences of myelopathy with fractional doses of 3 Gy or less (Table 1). Based on these and other series using larger doses per fraction, the incidence of myelopathy at 45 Gy in fractions of 1.8-2.0 Gy is most likely below 0.2% and is certainly less than the 5% quoted in major text books. Our best estimate of the conventionally fractionated dose causing a 5% incidence is 5761 Gy (9, 11, 14); and for 50% incidence, the dose is probably in the 68-73 Gy range (4, 16, 17, 18). Unfortunately, the published clinical data are inadequate for valid statistical dose-response analysis ( 18). The rare myelopathies that do occur at low doses (145 Gy) are seen for three reasons: (a) extrinsic factors reduce some individual’s radiation tolerance; (b) tens of thousands of patients are irradiated annually at these doses; and (c) the true dose was larger than estimated. The dogma that 45 Gy is associated with a 5% spinal cord complication rate ( 10) is clearly incorrect. Because 45 Gy conventionally fractionated is on the tail of the dose-response curve, variations around this dose will not appreciably alter the incidence of myelopathy. Thus, a difference in response between 40 Gy and 45 Gy or between schedules using 1.5 and 2.0 Gy per fraction would be undetectable. An idea that has been established mainly by tradition concerns the relative tolerance of the thoracic and cervical levels of the spinal cord. The dogma is that the thoracic cord is more sensitive than the cervical cord. This is attributed to the “poor vascular supply” (12) as evidenced by fewer radicular arteries, a narrowing of the ventral artery, and fewer central arteries. However, in answer to this concept Gillilan (7) states explicitly that “The blood supply to the thoracic cord is entirely adequate . . . and it is relatively as good as for any other cord segment.” Furthermore, the distribution of radiation lesions in the spinal cord is more typical of a venous lesion than an arterial one (8). The probable origin for the belief in a difference in sensitivities of the cervical and thoracic levels is the effect of the one-field-per-day technique used until the mid 1970’s. In a seminal paper entitled “Radiation Tolerance