To update a combined analysis of all published clinical data.

We collected data from 38 additional patients treated in our department or published in four different reports and calculated the biologically effective dose (BED) according to the linear-quadratic model using an alpha/beta value of 2 Gy for cervical and thoracic cord and 4 Gy for lumbar cord. In this model, a dose of 50 Gy given in single daily fractions of 2 Gy is equivalent to a BED of 100 Gy(2) or 75 Gy(4).

The 2005 risk score based on three variables (cumulative BED, highest BED of all treatment series in a particular individual, and interval), which discriminate three different risk groups, does not require modification. The low-risk group now contains 1 case of radiation myelopathy (RM) after hypofractionated stereotactic reirradiation. Therefore, the rate increased from 0% to 3%. Intermediate-risk patients developed RM in 25%, and high-risk patients in 90%. When the interval between the two treatment courses is not shorter than 6 months and the dose of each course is < or =98 Gy(2), the cumulative BED where no case of RM has yet been reported is 120 Gy(2).

Based on these updated results, the risk of RM appears small after < or =135.5 Gy(2) when the interval is not shorter than 6 months and the dose of each course is < or =98 Gy(2). We would recommend limiting the dose to the lowest feasible level. The influence of very steep dose gradients from stereotactic and intensity-modulated approaches (i.e., a more complex volume-effect) requires further evaluation.