Results of a Single Institution Experience with Dose-Escalated Chemoradiation for Locally Advanced Unresectable Non-Small Cell Lung Cancer

Background We determined factors associated with morbidity and outcomes of a series of non-small cell lung cancer (NSCLC) patients treated with dose-escalated chemoradiotherapy at the University of Pittsburgh Lung Cancer Program. Methods and materials The records of 170 stage III NSCLC patients treated with definitive intent were retrospectively reviewed. All patients received four-dimensional CT simulation scan and had respiratory gating if tumor movement exceeded 5 mm. Overall survival (OS), locoregional control (LRC), and freedom from distant metastasis (FFDM) were calculated using log-rank and Cox regression analysis. Results For the present series of patients, median follow-up was 36.6 months, median survival 27.4 months, and the 2- and 4-year OS was 56.0 and 30.7%, respectively. The 4-year LRC and FFDM were 43.9 and 40.7%, respectively. No benefit was associated with irradiation doses above 66 Gy in OS (p = 0.586), LRC (p = 0.440), or FFDM (p = 0.230). On univariate analysis, variables associated with worse survival included: clinical stage IIIB (p = 0.037), planning target volume (PTV) over 450 cc (p < 0.001), heart V30 over 40% (p = −0.048), and esophageal mean dose over 20% (p = 0.024), V5 (p = −0.015), and V60 (p = −0.011). On multivariable analysis, PTV above 450 cc (52.2 vs. 25.3 months, p < 0.001) and esophageal V60 >20% (43.8 vs. 21.3 months, p = −0.01) were associated with lower survival. Grade 2 or higher acute lung toxicity and esophagitis were detected in 9.5 and 59.7%, respectively of patients. Grade 2 or higher acute lung toxicity was reduced if lung V5 was ≤65 (7.4 vs. 23.8%, p = 0.03). Grade 2 or higher acute esophagitis was reduced if V60 ≤ 20% (62 vs. 81.3%, p = 0.018). The use of intensity-modulated radiation therapy was more frequent in stage IIIB compared to stage IIIA patients (56.5 vs. 39.5%, p = 0.048) and was associated with a higher lung V5 and V10. Conclusion The outcomes of a program of dose-escalated chemoradiotherapy for unresectable stage IIIA and IIIB NSCLC patients were consistent with other studies and showed no benefit to radiation doses above 66 Gy. Furthermore, maintaining low esophageal V60 and lung V5 were associated with lower morbidity and mortality.

conclusion: The outcomes of a program of dose-escalated chemoradiotherapy for unresectable stage IIIA and IIIB NSCLC patients were consistent with other studies and showed no benefit to radiation doses above 66 Gy. Furthermore, maintaining low esophageal V60 and lung V5 were associated with lower morbidity and mortality. The optimal chemoradiation therapy management of locally advanced non-small cell lung cancer (NSCLC) has been a subject of great interest (1). Even with current aggressive multi-modality treatment protocols, the clinical outcomes remain suboptimal (2). Previous clinical trials have established the superiority of concurrent chemoradiation using radiation doses between 60 and 66 Gy and revealed median survival times of 16-18 months (3)(4)(5). Dose-escalation protocols in Phase II clinical trials revealed the feasibility of radiation doses up to 74 Gy and suggested some increase in median survival to 26 months (6). Advances in radiation therapy techniques such as intensity-modulated radiation therapy (IMRT), three-dimensional conformal radiation therapy (3D-CRT), motion management, and attention to normal tissue dose parameters allowed dose escalation in some studies (7,8).
Radiation dose-escalation trials were evaluated for improved locoregional control (LRC) and overall survival (OS). A recent clinical Trial from the RTOG (RTOG 0617) for stage IIIA/B NSCLC (2) was initiated to randomize radiation dose groups to cohorts of 60 Gy compared to 74 Gy each with similar concurrent chemotherapy. The 74-Gy arm showed no benefit and was associated with a survival detriment. Both arms had the same LRC, distant metastasis rate and measured toxicity; however, further analysis showed that the use of IMRT was associated with better patient reported quality of life, and also that improved baseline pretreatment QOL was predictive of survival (9). Other parameters influencing survival included low heart dose, low esophageal morbidity, and smaller planning target volume (PTV). Although, it is important to note, the lack of pretreatment quality assurance from a central review has called the outcomes in question, especially when it comes to delineating normal structures such as the heart (10,11).
We now report the results of this single institution experience with definitive chemoradiation for stage IIIA and IIIB NSCLC and compare outcomes for patients treated with 3D-CRT compared to patients treated with IMRT.

Patient Population
We carried out a retrospective study using the criteria established by the American Joint Committee on Cancer (AJCC) seventh edition for patients with Stage IIIA and IIIB NSCLC. All patients were treated with definitive intent over the interval 2001-2013. All patients were determined to be unresectable by either: multiple node positivity, contralateral or supraclavicular lymph nodes, or tumor invasion of adjacent organs.
Patient medical records were de-identified and analyzed with ethical approval by the University of Pittsburgh Institutional Review Board (IRB #PRO13020306). The review was done with compliance to our ethic standards. All patients received daily radiotherapy fractions using a megavoltage linear accelerator with photon energy above 6 MV, and either IMRT or 3D-CRT. Radiation therapy was delivered concurrently with dual agent chemotherapy. All patients received four-dimensional CT (4D-CT) scan to assess target motion at time of simulation. Respiratory gating was included in daily treatments if tumor movement during simulation was greater than 5 mm in any direction (12). staging The staging system for NSCLC changed during the years of the study, therefore, we used the AJCC seventh staging edition as a standard for all patients. Patients were staged using PET/CT of the chest, abdomen, and pelvis, brain MRI, mediastinal lymph node sampling, and biopsy of both primary cancer and multiple mediastinal lymph nodes.

Dosimetry Parameters
In general, the normal esophagus was defined as being from the cricopharyngeus muscle to the gastroesophageal junction and the lung volume including the bilateral lungs from the apex chest apex to the diaphragm while subtracting out the PTV. These were mostly already created using our treatment plan software and we also used paper charts to assist with determine dosimetry values. In general our goals were V20 for lung <30%, mean esophagus dose less than or equal to 35 Gy and cord dose less than or equal to 45 Gy. Heart dose varied based on tumor location. For IMRT we also added V5 <60% and V10 <40% constraints for lung. In our early experience, patients were treated with fixed-field IMRT and after expanded to both rapid arc and fixed-field IMRT. statistical analysis LRC, freedom from distant metastasis (FFDM), and OS were primary outcome measures, and were calculated using log-rank and Cox regression analysis from time of diagnosis. Locoregional control was defined as lack of progression of clinical disease, seen on follow-up imaging or biopsy of the radiation therapy treatment volume for primary and regional lymph nodes. Acute toxicity was evaluated using binomial regression and late toxicity evaluated using Cox regression, t-test, ANOVA, binomial regression, and linear regression analysis. Each outcome measurement was correlated to dosimetric variables. Multivariable analysis (MVA) was performed for each factor found to be significant on unvariant analysis (p ≤ 0.05). Statistical analysis was performed using IBM SPSS Statistics Version 23.

Patient Demographics
There were 597 patients with clinical stage IIIA or IIIB NSCLC in the UPCI database over the interval 2001 to 2013. Patients treated with palliative intent (n = 296) were excluded as were those receiving only surgical management (n = 28). Radiotherapy patients not treated with concurrent chemotherapy (n = 39) were excluded. The remaining cohort of 213 patients was evaluated. Forty-three patients were lost to follow-up defined as no follow-up from treatment. Median follow-up of the cohort of 170 patients was 22 months and is shown in Table 1. The Interquartile Range (IQR) was 8.9-39.6. Median follow-up was

Treatment and Dosimetric evaluation
All patients were treated with concurrent chemoradiation, median radiation dose was 72 Gy (IQR, 68-77). The treatment volume and doses had inhomogeneity corrections. IMRT was used in 27.1%. All patients received a 4D-CT simulation scan, and 13.5% required respiratory gating. Concurrent carboplatin and paclitaxel were the dual agents in 88.2% of patients.

Overall survival
The

Factors influencing Toxicity
In the present series, the median value of mean lung dose was 15.2 Gy (IQR, 13-18 Gy) and median lung V20 was 26% (IQR 21-31%). There was grade 2 or higher acute lung toxicity detected in 9.5% of patients. On univariate analysis, the use of 3D-CRT (p = 0.036) and V5 ≤ 65 (p = 0.036) were associated with a lower rate of pneumonitis. On multivariate analysis, lung V5 ≤ 65 was associated with a decreased acute grade 2 or higher lung toxicity (7.4 vs. 23.8%, p = 0.027). The incidence of 1-year and 2-year late grade 3+ pneumonitis was 1.2 and 2.1%, respectively. There were no detectable univariate factors that predicted for late grade 3 or higher pneumonitis.  The median value for the mean esophageal dose in the group was 28 Gy in the present study (IQR, 21-33 Gy). The median esophageal V60 was 16% (IQR, 3-27%). Acute grade 2 or higher esophagitis was detected 59.7% of patients. On univariate analysis, a lower radiation therapy dose as a continuous variable (p = −0.041) and esophageal V60 ≤ 20% (p = −0.027) were associated with lower rates of acute grade 2 or higher esophagitis. On multivariate analysis, esophageal V60 ≤ 20% was associated with a lower rate of acute grade 2 or higher esophagitis (62 vs. 81.3%, p = 0.018). We observed late 1-year and 2-year grade 3+ esophagitis in 4.5 and 6.5% of patients, respectively. There was no detectable univariate factor that predicted late grade 3 or higher esophagitis ( Table 3).

DiscUssiOn
Despite advances in delivery techniques for radiation therapy, patients with locally advanced NSCLC have displayed suboptimal outcomes, and median survival remains between 15 and 26 months (2,13). Dose-escalation protocols for radiation therapy doses above 60 Gy showed a benefit, but also revealed increased toxicities (14). The addition of concurrent chemotherapy to 60 Gy radiation resulted in further improved survival (3)(4)(5). While some phase II studies showed the feasibility of doseescalated chemoradiation therapy with higher survival outcomes, the recent RTOG 0617 study did not confirm a benefit to the use of higher radiation dose (2,6). We compared the RTOG 0617 results with those of our single institution series at UPCI. The median survival for the present cohort of 170 evaluable NSCLC stage IIIA and IIIB patients was 27.4 months. This value is comparable to the median survival reported in the recent RTOG 0617 trial. In both studies, better staging with PET/CT leading to stage migration and improvements in treatment delivery and better supportive care may have accounted for the improvement in survival (2).
The median radiation dose delivered to patients in our single institution study was 72 Gy. We determined the value of dose escalation by comparing those patients treated to doses above 66 Gy with those receiving doses below 66 Gy. This cut-off value was based upon prior clinical trials comparing sequential to concurrent chemoradiation (3)(4)(5). Prior trials showed that concurrent, dual agent, platinum-based chemotherapy was more effective than use of sequential single drugs in chemoradiation protocols (3)(4)(5). We found no benefit for those patients treated above 66 Gy. We also found no benefit if the radiation dose was analyzed as a continuous variable. Therefore, we conclude that our data correlate with the results of RTOG 0617 showing no benefit to radiation dose escalation above 66 Gy.
The present analysis revealed that PTV volumes of NSCLC tumor above 450 cc were associated with a lower OS, LRC, and FFDM, again confirming the results presented in RTOG 0617 (2). Another clinical trial, which is a randomized phase II program, which evaluates adaptive planning in stage IIIA/B NSCLC patients for dose escalation, RTOG 1106, is ongoing. The control arm in this new study will receive 50 Gy, then continuing to 60 Gy after an interim PET/CT scan. The experimental arm will receive 46.2 Gy, then PET/CT scan evaluation, then use adaptive doseescalation ranges using the FDG-PET/CT scan up to a total dose of 80.4 Gy.
Our study revealed new information on toxicity of chemoradiotherapy of NSCLC. Acute grade 2 or higher lung toxicity in 9.5% patients was associated with a median lung V20 of 26 Gy. The data correlate well with a prior study showing that V20 values between 22 and 31 Gy led to an 8% pneumonitis rate (8). In this prior study (8), 42% of patients received chemotherapy (concurrent or sequential). In our study, the low rate of pneumonitis with high dose concurrent chemoradiation was likely attributable to 4D-CT simulation and the pre-screening of patients with tumor movement to include respiratory gating in their treatment program. Lower rates of lung toxicity were also detected in patients, where there was V5 ≤ 65%. A prior study (15), showed that V5 ≤ 42% did indeed correlate with a lower rate of acute pneumonitis for NSCLC patients treated with concurrent chemoradiation (15). A review of 220 esophageal cancer patients (16) also treated with radiation alone showed that V5 ≤ 60% correlated with lower rates of acute pneumonitis. We conclude that mean lung dose and lung V20 are the standard dosimetric parameters for predicting pneumonitis; however, we suggest that more attention should be given to the low dose volumes when treating patients with IMRT.
Esophagitis remains a serious complication of chemoradiotherapy of NSCLC. Esophageal grade 2 or higher toxicity rate was observed in 59.7% of our patients. This result was consistent with the combined grade 2 or higher esophagitis rate of the 74 Gy arm (43%) in the RTOG 0617 study (2). Keeping the esophageal V60 ≤ 20% correlated with lower frequency of detection of acute grade 2 or higher esophagitis. A retrospective analysis of another series of 109 NSCLC patients treated with concurrent chemoradiation showed the esophageal V60 correlated with higher rates of acute esophagitis (17). In yet another study, the threshold dose of 58 Gy predicted acute grade 3 or higher esophageal toxicity (18). Therefore, there is a consensus that the magnitude of the esophageal volume receiving more than 40-50 Gy correlates with the severity of acute esophagitis (19,20). Our results support efforts to ensure that as much of the esophagus as possible should be spared from the high-dose region. We will plan on incorporating our knowledge of the esophageal V60 this into our treatment planning for stage III patients treated with definitive chemoradiation.
The RTOG 0617 study showed that the use of IMRT did improve the QOL in NSCLC patients and reduced rates of grade 3 pneumonitis. In addition, there was an increased likelihood of completing adjuvant chemotherapy (2,9). We found that IMRT use was greater for IIIB patients, likely due to the requirement for coverage of supraclavicular and contralateral mediastinal disease, while achieving acceptable V20 volumes (mean, 30%) and attempting to still reduce lung toxicity. IMRT treatment plans decreased the esophageal V60 volume and the maximum heart dose. There was a trend for IMRT treatment plans to decrease the heart V30 volume, a parameter associated with survival outcomes on univariate analysis, but not multivariate analysis. A lowered heart V30 was also associated with improved survival outcomes in the RTOG 0617 study (2).
There was no statistically significant increase in survival for IIIA or IIIB patients treated with IMRT vs. 3D-CRT, although median survival times were different. In a secondary analysis of the data in RTOG 0617, the high-dose arm was associated with a lower patient reported QOL at 3 months, and baseline pretreatment QOL was predictive of survival (9). IMRT use was associated with a less deterioration of QOL compared to patients receiving 3D-CRT. We were unable to associate use of IMRT with a possible survival benefit, although a trend in this direction was apparent. A larger study with a more balanced distribution between uses of IMRT compared to 3D-RT may result in a statistically significant difference.
Patients treated with IMRT did not show a decreased rate of acute esophagitis or pneumonitis. This result may have been attributable to a higher usage of IMRT for patients with IIIB disease, and the need for larger treatment volumes. Two retrospective reviews of NSCLC patients showed decreased rates of pneumonitis with use of IMRT when compared to 3D-CRT for NSCLC (7,21). Radiation therapy planning comparing 3D-CRT to IMRT showed that IMRT also lead to a 50% relative reduction in predicted esophageal complications (22).
We analyzed our data using Kaplan-Meier plots, and this method revealed separation of the LRC and FFDM curves from OS (Figure 1). The data suggest that patients in our cohort may have died from causes other than local or distant recurrence of NSCLC. Since patients with aerodigestive cancers have a 3-5% yearly risk of a second aerodigestive primary cancer, we recommend that radiation oncologists ensure that long-term follow-up visits include screening for second cancers.
There were limitations to our retrospective analysis of NSCLC patients, including: (1) single institution retrospective review, (2) lack of a prospective randomized trial, and (3) limited follow-up interval. These factors might have led to both overestimating the primary outcomes and underestimating late toxicity. We also have a limited number of patients, and thus, our reports should be compared with other single institution reports and randomized control trials. We are additionally limited since we did not have electronic records for most of our patients and, therefore, may lead unreported events. Last, our dosimetry data are based on treating physicians who may have used different treatment considerations. We have overcome this issue by implementation of clinical pathways to guide our physicians and to have a more uniform criterion for care.
In a retrospective review of stage IIIA/B, NSCLC patients at UPCI larger PTV volumes and esophageal V60 > 20% were associated with poorer survival. Acute grade 2 or higher lung toxicity was lower in the V5 ≤ 65% group and acute grade 2 plus esophagitis was lower in the V60 ≤ 20% group. We recommend that patients with unresectable stage IIIA and IIIB NSCLC, who require radiotherapy be treated to doses not exceeding 66 Gy.
aUThOr cOnTriBUTiOns MB performed the retrospective review of the data analyzed in the manuscript and performed the majority of the statistics and was responsible for drafting and editing the manuscript. SG assisted with statistics and drafting of the manuscript. BG assisted in drafting the manuscript, statistical analysis, and literature review. SB oversaw drafting of the manuscript and edits and outlined the message and topic. DH assisted in drafting the manuscript and providing edits. JL provided edits concerning surgical outcomes. DF and MS reviewed the manuscript for medical oncology. JG led the team and edited the manuscript.