In-vivo dosimetry by diode semiconductors in combination with portal films during TBI: reporting a 5-year clinical experience.

Background and purpose: In-vivo dosimetry is vital to assure an accurate delivery of total body irradiation (TBI). In-vivo lung dosimetry is strongly recommended because of the risk of radiation-induced interstitial pneumonia (IP). Here we report on our 5-year experience with in-vivo dosimetry using diodes in combination with portal films and assessing the effectiveness of in-vivo dosimetry in improving the accuracy of the treatment. Moreover, we wished to investigate in detail the possibility of in-vivo portal dosimetry to yield individual information on the lung dose and to evaluate the impact of CT planning on the correspondence between stated and in-vivo measured doses. Materials and methods: From March 1994 to March 1999, 229 supine-positioned patients were treated at our Institute with TBI, using a 6 MV X-rays opposed lateral beam technique. 146 patients received 10 Gy given in three fractions, once a day (FTBI), shielding the lungs by the arms; 70 received 12-13.2 Gy, given in 6-11 fractions, 2-3 fractions per day (HFTBI): in this case about 2/3 of the lungs were shielded by moulded blocks (mean shielded lung dose equal to 9 or 9.5 Gy). Thirteen patients received 8 Gy given in a single fraction (SFTBI, lung dose: 7 Gy). For all HFTBI and FTBI patients, midline in-vivo dosimetry was performed at the first fraction by positioning two diodes pairs tone at entrance and one at the exit side) at the waist (umbilicus) and at the pelvis (ankles). If at least one of the two diodes doses (waist-pelvis) was outside +/-5% from the prescribed dose, actions could be initiated, together with possible checks on the following fractions. Transit dosimetry by portal films was performed for most patients; for 165 of them (117 and 48, respectively for FTBI and HFTBI) the midline in-vivo dose distribution of the chest region was derived and mean lung dose assessed. As a CT plan was performed for all HFTBI patients, for these patients, the lung dose measured by portal in-vivo dosimetry was compared with the expected value. Results: Concerning all diodes data, 528 measurements were available: when excluding the data of the first fraction(s) of the patients undergoing corrections (n = 392), mean and SD were respectively 0.0% and 4.5% (FTBI: -0.3 +/- 4.8%; HFTBI: 0.4 +/- 3.9%). In total 105/ 229 patients had a change after the first fraction and 66/229 were controlled by in-vivo dosimetry for more than one fraction. Since January 1998 a CT plan is performed for FTBI patients too: when comparing the diodes data before and after this date, a significant improvement was found (i.e. rate of deviations larger than 5% respectively equal to 30.7% and 13.1%, P = 0.007). When considering only the patients with a CT plan, the global SD reduced to 3.5%. Concerning transit dosimetry data, for FTBI, the mean (midline) lung dose was found to vary significantly from patient to patient (Average 9.13 +/- 0.81 Gy; range 7.4-11.4 Gy); for the HFTBI patients the mean deviation between measured and expected lung dose was 0.0% (1 SD = 3.8%). Conclusions: In vivo dosimetry is an effective tool to improve the accuracy of TBI. The impact of CT planning for FTBI significantly improved the accuracy of the treatment delivery. Transit dosimetry data revealed a significant inter-patient variation of the mean lung dose among patients undergoing the same irradiation technique. For patients with partial lung shielding (HFTBI), an excellent agreement between measured and expected lung dose was verified. (C) 1999 Elsevier Science Ireland Ltd. All rights reserved.