The medical records and dose volume histogram (DVH) data of 60 lung cancer patients, treated with three dimensional (3D) conformal RT from August 2005 to July 2006 at Seoul St. Mary's Hospital, were retrospectively collected and analyzed. At least 6 months of follow-up was required for inclusion in this study. Patients who died within 6 months after completion of RT, were lost to follow-up, or did not complete treatment, were excluded from the analysis.

Patient characteristics are summarized in Table 1. All 60 patients were treated with curative intent, using 3D conformal RT. Forty-eight patients (80%) received chemotherapy as a part of their definitive treatment. Twenty-seven patients (45%) received chemotherapy prior to RT and 21 patients (35%) received concurrent chemoradiotherapy. The regimens used for concurrent chemotherapy were etoposide and cisplatin (8 patients), docetaxel and cisplatin (8 patients), taxol (4 patients), and cisplatin (1 patient). Fourteen patients (23.3%) received RT or concurrent chemoradiotherapy as definitive salvage treatment following recurrent disease after surgical intervention (lobectomy eight patients, wedge resection two patients, pneumonectomy one patient, and others three patients). Fifty-two patients (86.7%) showed a good performance status of more than Eastern Cooperative Oncology Group (ECOG) 2. Each patient’s pre-RT chest CT scans were reviewed by two radiation oncologists and one diagnostic radiologist who were not informed of which patients developed symptomatic RP. Fifteen patients (25%) had interstitial changes on their chest CT scans. Interstitial changes were recorded if any of the following criteria were met: subpleural ground glass opacity (GGO), fibrous reticulation, traction bronchiectasis, micronodules, honey-comb appearance, or volume loss, especially in both lower lobes [18]. Thoracic RT was not performed in patients with clinical ILD, which was defined as a status of symptomatic disease or post-treatment with high probability of acute exacerbation. The patients showing interstitial change on pre-RT CT without related symptom or history of a treatment were included in this study as subclinical ILD. This retrospective study was approved by the hospital’s institutional review board.

A planning CT scan with contrast enhancement of the entire lung volume was acquired with a vac-lock immobilization device at shallow normal breathing. Four dimensional CT (4D-CT) scans were also acquired in all patients to evaluate respiratory motion. All patients underwent virtual simulation and treatment planning with the 3D treatment planning system (Pinnacle, ADAC Laboratories, Milpitas, CA). The calculation was performed with lung heterogeneity corrections by radiological path algorithm. The gross tumor volume (GTV) included the tumor and the clinically involved nodes. The GTV node was determined by chest CT and/or positron emission tomography–computed tomography for the majority of patients; the node was determined by mediastinoscopy in a small number of patients. The clinical target volume included the GTV with an additional uniform 5-mm expansion and high risk, uninvolved mediastinal lymph nodes as well as the ipsilateral hilar lymph nodes. The high risk, involved mediastinal lymph nodes were defined as those lymph nodes adjacent to the GTV node with a high probability of microscopic involvement. We added a different panning target volume (PTV) margin based on an assessment of each tumor’s motion using the Leonardo system (Siemens, Concord, CA) in 4D-CT. The lung, esophagus, spinal cord, and heart were contoured. The lung volume was contoured and defined excluding the GTV.

Most patients were treated with conventional beam arrangements (i.e., initial AP/PA fields to include the PTV) for the first 30-40 Gy, followed by an off-cord oblique beam to the PTV, usually composed of three to five beams. Some patients with small PTVs were treated with multiple beam arrangements from the beginning. The median radiation dose was 56.9 Gy (range, 45.0 to 64.8 Gy), generally delivered at 1.8-2 Gy/fraction, five fractions per week. RT was delivered with 10 MV X-ray from a LINAC (Siemens). Generally, the radiation dose was prescribed such that 95% of PTV was covered by 95% of the prescription dose. The lung DVH was calculated directly from the physical dose distribution (i.e., adjustments were not made for fraction size or overall treatment time). The following dosimetric factors were extracted from the lung DVH: V₁₀, V₂₀, V₃₀, V₄₀, mean lung dose (MLD), and normal tissue complication probability (NTCP) as derived from the Lyman and Kutcher models.

A diagnosis of RP was made by the clinical symptoms of nonproductive cough, fever, exertional dyspnea and characteristic chest radiographs and CT findings. Patients were evaluated by radiation oncologists 3 to 6 weeks after completion of RT and at 1- to 3-month intervals thereafter. RP grading was recorded using Radiation Therapy Oncology Group (RTOG) scale, based on the severity of clinical symptoms and radiographic changes. Two radiation oncologists and one diagnostic radiologist who were blinded to RP status reviewed the medical records, chest X-ray and/or chest CT scan of all patients with suspected RP, and ruled out other causes of pneumonia that were confirmed as infectious pneumonia or disease progression.

All statistical analysis in this study was performed with the SPSS ver. 18.0 (SPSS Inc., Chicago, IL). The chi-squared test and Fisher exact test were used to compare RP risk in subgroups and Student’s t test was used for normally distributed continuous variables. To identify independent predictors of RP, all factors were included in a multivariable linear regression model. Using a stepwise procedure, variables with a p-value of 0.05 or higher were dropped from the model so that only significant predictors remained. All statistical tests were two sided and a p-value of less than 0.05 was considered statistically significant.

With a median follow-up of 15.5 months (range, 6.1 to 40.9 months), 22 patients (36.7%) developed RP: eight (13.3%) with grade 1, six (10.0%) with grade 2, one (1.6%) with grade 3, 2 (3.3%) with grade 4, and five (8.3%) with grade 5. RTOG grade 2 or higher RP was observed in 14 patients (23.3%). Five patients (8.3%) died from RP. The median time to the onset of symptoms was 33 days (range, 6 to 160 days) from the completion of RT.

Table 2 shows the relationship between clinical factors and RP. There was a trend of increased risk of RP in patients treated with combined chemoradiation compared with patients treated with RT alone (41.7% vs. 16.7%, p=0.108). However, this difference was not statistically significant. Interstitial change in the pre-RT chest CT was the only clinical factor associated with RP (p=0.033). In multivariable analysis, interstitial changes in the pre-RT chest CT were significantly associated with RP (p=0.009; odds ratio, 10.24; 95% confidence interval, 1.787 to 58.679).

The statistically significant dosimetric factors for predicting RP are shown in Table 3. In the univariable analysis, the radiation field size, MLD, V₁₀, V₂₀, V₃₀, V₄₀, and NTCP were significantly associated with RP (all p < 0.05). From the multivariable analysis, the MLD (p < 0.001) and V₃₀ (p < 0.001 were significant independent predictors for RP. Cumulative curves of the development of RP stratified by MLD, V₂₀, V₃₀, and NTCP are illustrated in Fig. 1. Our data showed that maintaining an MLD less than 15 Gy, V₂₀ ≤ 25 %, and V₃₀ ≤ 20 % could reduce the incidence of RP. Fig. 2 illustrates the distribution of RP grades by MLD, V₂₀, V₃₀, and NTCP. As shown in Fig. 2, five patients with grade 5 RP did not have higher dosimetric parameters compared to non-fatal RP patients.

Fifteen patients (25%) had interstitial changes on their chest CT scans. There was a difference in RP incidence between patients with interstitial lung change in pre-RT CT and without interstitial change (Table 4). RP was detected in nine (60.0%) of the 15 patients with diffuse interstitial change in their pre-RT CT compared to 13 (28.9%) of 45 patients without interstitial changes (p=0.030). Risk of RP grade ≥ 2 was significantly different depending on presence of interstitial change in the patient’s pre-RT CT (p=0.030). In addition, interstitial change in pre-RT CT raised the risk of RP grade ≥ 3 and ≥ 4 from 4.4% to 40% (p=0.002) and 4.4% to 33.3% (p=0.008). Interstitial change in pre-RT chest CT was the only clinical factor associated with RP in the multivariable analysis (p=0.009).

We analyzed the pre-RT chest CT of the RP patients, by focusing on the imaging findings. Of these patients, the characteristic findings of the nine RP patients with diffuse interstitial changes in their pre-RT chest CT scans are summarized in Table 5. These patients presented mixed and various interstitial changes such as honeycombing, reticular opacity, traction brochiectasis, focal subpleural GGO, and micronodular density. However, in the higher grade RP patients, diffuse honeycombing and diffuse nodular density patterns were frequently observed involving the entirety of both lungs diffusely, or occasionally involving both lower lung fields. Fig. 3 shows a typical representative case of a patient who progressed to fatal RP. Four of the grade 5 RP patients had interstitial changes in their pre-RT chest CT and also received chemotherapy: concurrent chemoradiotherapy in two patients, and sequential chemotherapy followed by RT in two patients.

Table 6 gives a detailed description of five patients with grade 5 RP. Three patients initially developed focal consolidation corresponding to the radiation fields during RT or within one month after RT completion, and the lesions progressed to the entire lung beyond radiation fields. The other two patients presented with bilateral lung infiltration 4 to 6 weeks after RT completion and died following rapid progression. No pathogens were identified and steroid and antibiotic treatments were not effective.