Send to

Choose Destination
Int J Radiat Oncol Biol Phys. 2019 Feb 1;103(2):403-410. doi: 10.1016/j.ijrobp.2018.09.029. Epub 2018 Oct 3.

Validation of Effective Dose as a Better Predictor of Radiation Pneumonitis Risk Than Mean Lung Dose: Secondary Analysis of a Randomized Trial.

Author information

Department of Radiation Oncology, The University of Texas MD Anderson Cancer Center, Houston, Texas.
Department of Radiation Oncology, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts.
The D-Lab: Decision Support for Precision Medicine, GROW-School for Oncology and Developmental Biology, Maastricht University Medical Centre, The Netherlands.
Department of Radiation Oncology, University of Nebraska Medical Center, Omaha, Nebraska.
Department of Radiation Physics, The University of Texas MD Anderson Cancer Center, Houston, Texas.
Department of Radiation Oncology, The University of Texas MD Anderson Cancer Center, Houston, Texas. Electronic address:



To confirm the superiority of effective dose (Deff) over mean lung dose (MLD) for predicting risk of radiation pneumonitis (RP), using data from patients on a randomized trial of intensity modulated radiation therapy (IMRT) versus passively scattered proton therapy (PSPT).


The prescribed target dose for the 203 evaluated patients was 66 to 74 Gy (relative biological effectiveness) in 33 to 37 fractions with concurrent carboplatin/paclitaxel. Time to grade ≥2 RP was computed from the start of radiation therapy, with disease recurrence or death considered censoring events. Generalized Lyman models of censored time to RP were constructed with MLD or Deff as the dosimetric parameter. Smoking status (current, former, never) was also analyzed.


Of the 203 patients, 46 experienced grade ≥2 RP (crude incidence 23%) at a median 3.7 months (range, 0.6-12.6 months). The volume parameter estimated for the Deff model was n = 0.5, confirming estimates from earlier studies. Compared with MLD (in which n = 1), the dosimetric parameter Deff, computed using n = 0.5, resulted in a better fit of the Lyman model to the clinical data (P = .010). Using Deff, the model describes RP risk for IMRT and PSPT data combined because no further improvement was found from separate fits (P = .558). Based on Deff, predicted RP risk per patient ranged from 24 percentage points lower to 19 percentage points higher than predictions based on MLD. For patients with similar MLD, Deff predicted higher risk, on average, for PSPT over IMRT. Current smokers had a lower risk of RP compared with former smokers and nonsmokers (P = .021).


We used data from a randomized trial to validate our previous finding that Deff with n = 0.5 (corresponding to root mean squared dose) is a better predictor of RP than is MLD. Differences between Deff and MLD indicate that delivering higher doses to smaller lung volumes (vs lower doses to larger volumes) increases RP risk. We further corroborated that current smoking is associated with decreased RP risk.

Supplemental Content

Full text links

Icon for Elsevier Science
Loading ...
Support Center