Patterns of Care and Outcomes of Intensity-Modulated Radiotherapy and 3D Conformal Radiotherapy for Early Stage Glottic Cancer: A National Cancer Database Analysis

Early stage glottic cancer has traditionally been treated with 3D conformal radiotherapy (3DCRT). However, intensity-modulated radiotherapy (IMRT) has been recently adopted as an alternative to decrease toxicity. Here, we compared the usage and outcomes of IMRT and 3DCRT for patients with early stage squamous cell carcinoma (SCC) of the glottic larynx. Using the National Cancer Database, we identified patients with Stage I–II SCC of the glottis who received 55–75 Gy using IMRT (n = 1623) or 3DCRT (n = 2696). Median follow up was 42 months with a 5-year overall survival (OS) of 72%. Using a nominal logistic regression, race, ethnicity, year of diagnosis and fraction size were associated with the receipt of IMRT (p < 0.05). Using Kaplan–Meier methods and Cox proportional hazards models as well as a propensity matched cohort, there was no difference in OS for patients who received IMRT versus 3DCRT (hazard ratio (HR), 1.08; 95% confidence interval (95% CI), 0.93–1.26; p = 0.302). However, there was a survival benefit for patients receiving slight hypofractionation as compared to conventional fractionation (HR, 0.78; 95% CI, 0.69–0.92; p = 0.003). IMRT was associated with similar survival as 3DCRT, supporting the implementation of this potentially less toxic modality without compromising survival.


Introduction
Laryngeal cancer has a global incidence of 238,150 cases per year with glottic cancer representing approximately one-half of all laryngeal cancers [1,2]. Early stage squamous cell carcinoma (SCC) of the glottis is often treated with radiotherapy (RT) alone [3] and has traditionally been treated with 3D conformal radiotherapy (3DCRT) with excellent rates of long-term local control [4,5]. Generally, patients with early stage glottic SCC have a low rate of acute and late treatment toxicities. Nevertheless, patients are at risk to develop severe long-term side-effects of dysphagia, laryngeal/soft tissue necrosis and carotid artery stenosis leading to cerebrovascular events [4,[6][7][8][9][10][11]. Recently, intensity-modulated radiotherapy (IMRT) has been adopted as an alternative modality to decrease toxicity, including strategies such as reducing dose to the carotid arteries and radiating a single vocal cord [12][13][14][15]. The goal of IMRT in glottic cancer is to prevent long-term toxicity without compromising already  Figure 1 depicts the utilization of IMRT and 3DCRT by year of diagnosis.

Propensity Score-Matched Analysis
After matching for all covariates in Table 1 with p < 0.1, there remained no statistically significant difference in OS for patients treated with IMRT versus 3DCRT (HR, 0.90; 95% CI, 0.79-1.03; p = 0.127). Table 4 depicts the propensity score-match groups with no statistically significant difference between the groups. There was a total of 1428 matched cases included in the analysis with 714 records in each group.

Discussion
We used the NCDB to evaluate patterns of care for patients with early stage glottic SCC. We also estimated the OS differences between IMRT and 3DCRT. On both multivariable survival analysis and propensity score-matched analysis, there was no difference in OS for patients treated with IMRT as compared to 3DCRT. The goal of IMRT is to reduce toxicity without compromising treatment outcomes, and our findings support that OS is not compromised with IMRT.
Previous work has demonstrated that IMRT improves toxicity outcomes in head and neck cancer by, for example, reducing xerostomia [16][17][18]. In the setting of early stage glottic SCC, IMRT has the potential to reduce dose to the carotid arteries, which would potentially decrease the risk of stroke and more easily allow for future RT in the event of a second head and neck cancer diagnosis [4,[6][7][8][9][10][11]. Others have shown that clinical outcomes are not compromised with IMRT, and dosimetric analyses demonstrate significant dose reduction to the carotid arteries by up to 75% [4,13,19,20]. Mohamed et al. performed a retrospective case-control study comparing 3DCRT and IMRT, showing no statistically significant difference in oncologic outcomes (e.g., 3-year locoregional control and OS); however, there was also no statistically significant difference in the rate of cerebrovascular events (p = 0.7), feeding tube dependence (p = 0.5) or aspiration events (p = 0.4) [19]. Zumsteg et al. also performed a clinical comparison of IMRT and 3DCRT and found no difference in oncologic outcomes, with 3-year local control rates of 88% and 89%, respectively (p = 0.938), while maintaining low doses to the carotid arteries [13].
The goal of using IMRT for early stage glottic SCC is to reduce treatment-related toxicity. However, to our knowledge, there are no prospective data showing a decrease in toxicity for IMRT as compared to 3DCRT for patients with early stage glottic SCC. The NCDB does not provide data on toxicity from radiotherapy; thus we are unable to make conclusions on whether or not IMRT resulted in less toxicity as compared to 3DCRT. Despite lack of data demonstrating decreased toxicity with the use of IMRT, the use of IMRT has increased over recent years as shown in Figure 1. While some studies have utilized generous margins including the entire larynx plus 1 cm, new studies are investigating radiating a single vocal cord as opposed to the entire larynx [12,13]. Al-Mamgani et al. performed a retrospective analysis investigating the use of IMRT to a single, entire vocal cord for patients with T1a glottic cancer [12]. After a median follow-up of 30 months, the 2-year local control rate was 100%. There was no grade 3 toxicity, and there was grade 2 acute dermatitis or dysphagia in 17% of patients. These patients were compared to an older cohort of patients treated to the whole larynx with 3DCRT, though not all patients were treated with slight hypofractionation [21]. This rough comparison demonstrated comparable local control (p = 0.24) but more acute toxicity in the 3DCRT group (66% vs. 17%, p < 0.0001).
There is a clear survival benefit for patients with early stage glottic SCC treated with slight hypofractionation as opposed to conventional fractionation [22]. Our findings support this claim, as shown in Figure 3. Interestingly, we found that patients treated with IMRT were less likely to be treated with slight hypofractionation (OR, 0.56; 95% CI, 0.48-0.65; p < 0.0001). These observations suggest providers are more likely to use conventional fractionation when employing IMRT. Despite this finding, there was no survival difference in OS between IMRT versus 3DCRT overall (Figure 2, p = 0.251) or among patients who received 225 cGy per fraction (Figure 4, p = 0.342).
The current study is subject to the inherent limitations of a population-based retrospective study; however, it provides motivation to perform prospective studies comparing IMRT and 3DCRT for patients with early stage glottic SCC given the lack of clinical data. Furthermore, the NCDB dataset contains only survival data, which is a less informative endpoint to evaluate the oncologic results of radiotherapy techniques in patients with early-stage glottic cancers. Patient data for toxicity, local control and cancer-specific survival would better facilitate comparisons between IMRT and 3DCRT. Nevertheless, disease recurrence in early stage glottic cancers does impact survival [23], even if this impact is small. Consequently, the high patient numbers in the NCDB facilitate the detection of small differences in survival associated with different radiotherapy techniques. To this end, using the NCDB, we and others observed survival differences in patients treated with different fraction sizes which were not observed in the original randomized trials addressing this issue [22,[24][25][26]. Although the impact of IMRT on local control and toxicity is unknown, we did not find any difference in OS between IMRT and 3DCRT.

Study Design and Patient Data Set
We used the NCDB to perform a retrospective cohort study. The Commission on Cancer (CoC) of the American College of Surgeons and the American Cancer Society created the NCDB as a clinical oncology database sourced from CoC-accredited hospital registry data, including approximately 70% of all new invasive cancer diagnoses in the US [27]. Collected data include demographics, comorbidities, tumor characteristics, staging details, surgical and adjuvant treatments (e.g., radiotherapy) and OS. At the time of analysis, patient data were available for cases diagnosed from 2004 through 2016.

Patient Selection
Patients with cT0-2, N0, M0 glottic SCC who were treated with definitive RT alone (3DCRT or IMRT) to a dose of 55-75 Gy were included in the analysis. Patients were excluded if they did not receive their first course of treatment at the reporting facility, were treated with definitive surgery, received chemotherapy or received immunotherapy. Figure 5 depicts the Consolidated Standards of Reporting Trials (CONSORT) diagram for inclusion and exclusion criteria.

Exposure Variables of Interest
Using the National Cancer Database, we identified patients with Stage I-II SCC of the glottis who received 55 to 75 Gy using IMRT (n = 1254) or 3DCRT (n = 3043). Patient, tumor and RT characteristics were stratified into groups. Patient age was organized into four groups (≤50, 51-60, 61-70 and >70 years of age). Comorbidity information was organized into three groups (0-1, 2 and ≥3 comorbidities) using the CDCS. Patient location of treatment was divided by US Census region. Population density of patient's residence was categorized as metropolitan, urban and rural. Patient distance from reporting facility was stratified into four groups (<10, 10-19, 20-29 and ≥30 miles). Patient county income and educational levels were categorized as quartiles based on equally proportioned ranges among all US zip codes. Education was defined as the percentage of adults in the patient's zip code without a high school diploma. When comparing fraction sizes, RT dose was dichotomized into 2 Gy fractions and 2.25 Gy fractions with all other fraction sizes being excluded.

Statistical Analysis
Variables associated with the receipt of IMRT or 3DCRT were assessed using nominal logistic regression. Overall survival (OS) was estimated using Kaplan-Meier methods and Cox proportional hazard models in the entire population and in a propensity matched cohort. Categorical variables were compared using chi-square tests. Continuous variables were compared using a two-sample t-test. Univariable and multivariable logistic regression were used to determine factors associated with the receipt of IMRT. Variables were included in the multivariable analysis if they were found to be associated with RT modality (p < 0.1) on univariable analysis. Overall survival (OS) was defined as the time from diagnosis to death. The Kaplan-Meier method and log-rank test were used to estimate and compare OS. Univariable and multivariable Cox proportional hazards modeling was used to determine factors associated with OS. Variables were included in the multivariable analysis if they were found to be associated with OS (p < 0.1) on univariable analysis. Sensitivity analysis using 1-to-1 propensity score matching was performed to ensure patients who received IMRT were equivalent to those who received 3DCRT with regard to all covariates. Groups were matched based on variables found to be statistically significant predictors of receipt of IMRT on univariate analysis. Analyses were performed using STATA MP 14 (StataCorp LLC, College Station, TX, USA) and SPSS (IBM Corporation, Armonk, NY, USA) statistical software. A p value less than 0.05 was considered statistically significant.

Conclusions
Although IMRT was associated with conventional fractionation, we observed similar survival between IMRT and 3DCRT, supporting the implementation of this potentially less toxic modality without compromising survival. A prospective comparison of IMRT and 3DCRT for patients with early stage glottic SCC is warranted.