br Although the Liao et al trial showed no

Although the Liao et al. trial showed no difference in primary endpoints, we caution against concluding that proton therapy has no benefit in NSCLC. Among patients who were not eligible for randomization, PSPT was the only modality that could meet dose constraints in 13 patients, illustrating that for subsets of patients with NSCLC, protons had a clear dosimetric advantage and treatment with IMRT may necessitate underdosing the tumor to meet normal tissue constraints. Moreover, patients in the cur-rent study were all treated with IMPT. A detailed dosimetric com-parison of IMPT, PSPT, and IMRT showed that IMPT could deliver the highest mean tumor dose (with improved conformality) while simultaneously achieving the lowest dose to heart, lungs, and A23187 for patients with anatomically complex thoracic tumors [20]. In our practice, patients with large bilateral tumors are usu-ally treated with IMPT because dose constraints are difficult to meet with PSPT or IMRT and because IMPT does not seem to com-promise local control. r> Efforts to escalate radiation dose with photon-based techniques to improve outcomes for patients with locally advanced NSCLC proved ineffective in the phase III randomized controlled trial RTOG 0617. That trial allowed treatment with IMRT or 3D confor-mal radiotherapy, and image guidance was encouraged but not mandated. Despite similar local control and cancer-specific sur-vival rates from standard-dose (60 Gy) and high-dose (74 Gy) ther-apy, overall survival was worse in the high-dose group. Several potential explanations have been advanced to explain the lack of benefit from dose escalation, among them that the high-dose
Fig. 1. Overall and local failure–free survival estimates by disease stage.
therapy may have caused more cardiopulmonary toxicity from higher doses to heart and lungs. This association suggests that sub-clinical cardiopulmonary toxicity was present but was not cap-tured by adverse event reporting. We also found a significant association between cardiac V40 and overall survival. Notably, we examined several other cardiac variables including mean and max heart dose as well as V10, V20, and V30; however, cardiac V40 showed the strongest association with overall survival. Because all of the cardiac dosimetric variables were codependent, only cardiac V40 was included in the multivariate model.
The detrimental effects of radiation dose to the heart and lungs have been established in numerous studies [21–23]. Speirs et al. 
analyzed heart and lung doses in 416 patients with locally advanced NSCLC and found that heart V50 and lung V5 were both independently associated with overall survival [24]. Similarly, Xu et al. measured cardiac troponin after concurrent chemoradiation and showed that mean heart doses >20 Gy led to significant increases in troponin and that a doubling or more in troponin levels relative to baseline was associated with worse overall sur-vival [25]. Although dose escalation was thought to be beneficial in several prospective trials before RTOG 0617, that benefit is likely negated if the dose escalation results in fatal cardiopulmonary tox-icity. Thus, even though RTOG 0617 did not show a benefit from dose escalation, its results are confounded by increased cardiopul-
Outcomes after IMPT for NSCLC
Table 2
Survival and disease control outcomes by disease stage.
1 Year 2 Years 3 Years
4 Years 5 Years
Overall survival
Disease-free survival (any failure)
Local–regional failure-free survival
Local failure–free survival
Regional failure–free survival
Distant metastasis-free survival
Table 3