A retrospective experience suggests that early glottic cancer treated with HRT results in a higher incidence of acute toxicity without improved control compared to CRT delivered in an appropriate OTT.
Michael T. Scott, MD, MBA
The optimal fractionation for early glottic cancer is unclear. Prospective randomized data showing superiority of 2.25 Gy versus 2 Gy/fraction derive from only 1 trial. We analyzed our single-institutional experience to address the therapeutic ratio of 2.25 Gy vs. 2 Gy.
Materials and Methods
One hundred patients with T1-T2 glottic carcinoma received radiation therapy (RT) between 2003-2013. Their records underwent retrospective analysis. Median age was 64 years. Seventy-six had T1 tumors; 24 had T2. Forty-eight received 2 Gy daily (conventional RT--CRT) and 52 received 2.25 Gy daily (hypofractionated RT--HRT). Median CRT dose was 66 Gy and HRT dose was 63 Gy. Logistic regression, t-test, log-rank test, and Cox proportional hazards regression were used and two-sided p-value was reported. Recurrence-free survival (RFS) was estimated using Kaplan-Meier method.
Median follow-up was 28.8 months overall, with 54.7 months for CRT and 16.1 months for HRT, after excluding from local control analysis 9 patients with follow-up <3 months. CRT patients had less grade 2 dysphagia (P = 0.014), grade >2 hoarseness (P = 0.001), grade >2 mucositis (P < 0.001), and required less frequent administration of opiates (P = 0.004). Rates of grade >2 radiation dermatitis (P = 0.421), grade >2 laryngeal edema (P = 0.549), and weight loss (P = 0.655) during treatment were similar. Complete response (CR) was 91.7% in the CRT group vs. 90.4% for HRT. RFS was comparable at 24 months (86.7% vs 86.6%, P = 0.988) and at 60 months (82.8% vs 78.8%, P = 0.655) in CRT and HRT patients. P-value of log-rank test for RFS was 0.797. Univariate analysis for the 76 T1 patients upheld the significance for all of the above findings, including CR and RFS, except grade 2 dysphagia (P = 0.156).
Our retrospective data suggest that treatment of early glottic cancer with 2.25 Gy/fraction may result in greater acute toxicity than 2 Gy/fraction without improved efficacy.
Early stage glottic cancers are a common malignancy of the head and neck, and may be treated effectively with radiotherapy (RT), surgery, or laser microsurgery.1 RT provides local control rates of 80-95%.2-10 Historically, conformal RT (CRT) used 2 Gy per fraction for 33 fractions. Some retrospective data have demonstrated that hypofractionated regimens (>2 Gy) with a shorter overall treatment time (OTT) improved local control. However, most reports used <2 Gy as the control.2-4, 11-13 Few retrospective studies have evaluated 2 Gy vs. 2.25 Gy.
* excluding 9 pts whose follow-up duration is less than 3 months
In 2006, Yamazaki published a prospective randomized trial demonstrating superior local control of 92% with HRT versus 77% with CRT.14 A limited discussion of toxicity indicated no significant differences in acute mucosal, skin or chronic adverse reactions. There are many differences in the techniques used in this study and in typical United States practice (beam energy, field size, dose prescription point, dose range). Biologic or cultural differences between a Japanese patient cohort and one from the United States may affect tumor control or toxicity outcomes. Since 2006, the Japanese randomized trial has not been replicated. For these reasons, the outcome and toxicity results from that study may not apply to patients treated in the United States at the University of Miami, we noted greater toxicity with HRT without apparent local control benefit compared to CRT. Therefore, we decided to re-examine this topic.Retrospective data collection and analysis was approved by our Institutional Review Board. One hundred patients with histologically proven T1-2N0M015 glottic carcinoma treated with RT at the University of Miami between 2003-2013 were identified. Patient and tumor characteristics are summarized in Table 1.Treatment was delivered with 6 MV photons to opposed lateral or slightly oblique fields. A few patients prior to 2006 also received an additional anterior field. In both eras, patients were treated to an isodose line (usually 98% or 95%) with fields ranging from 5x5 to 6x6 cm. Bolus was used on some anterior commissure cases in the CRT era but not in the HRT era.
Prior to 2006, patients were prescribed CRT daily to 66 Gy for T1 tumors or 70 Gy for T2 tumors. After 2006, patients were prescribed HRT daily to 63.25 Gy for T1 tumors or 65.5 Gy for T2 tumors. Forty-eight patients received 2 Gy and 52 received 2.25 Gy. Median dose was 66 Gy for CRT and 63 Gy for HRT. No patients received systemic therapy. Treatment characteristics are summarized in Table 2.SAS and R software were used for all statistical analyses (SAS Institute, Cary, NC and www.r-project.org). Logistic regression, t-test, log-rank test, and Cox proportional hazards regression were used to compare the two groups and two-sided P-value was reported. P <0.05 was considered statistically significant. Local control was defined at the primary site. Patients without CR at initial follow-up after radiotherapy were regarded local failures. Recurrence-free survival (RFS) was defined as no evidence of recurrence at any site and was estimated using Kaplan-Meier method.16 Complications were graded using CTCAE version 4.0.17Flexible laryngopharyngoscopy was performed at 6-12 weeks after radiotherapy to assess response. Rates of response and tumor progression were not statistically different for CRT and HRT groups (Table 3).
RFS rates were similar at 12, 24, 36 and 60 months as shown in Table 4. The p-value of log-rank test for RFS was 0.797 (Figure 1). To account for the variable stage distribution of patients in the CRT and HRT groups, a multivariate analysis adjusted for AJCC stage confirmed that RFS of the two treatment groups is not significantly different: HR (hazard ratio) 1.147 and (95% CI; P-value):0.012, 111.1; P =.953) by Cox proportional hazard regression. Univariate analysis for the 76 T1 patients upheld the significance for both CR rate and RFS. The 2-yr RFS was 87.3% for CRT and 88.9% for HRT (P = 0.825) (Figure 1). Cox proportional hazard regression showed that RFS of the two treatment groups are not significantly different: HR 0.382 and (95% CI; P-value) (0.001, 106.6; 0.738) for the T1 patients.There were 7 local failures in the CRT group, with 2/7 patients receiving less than the planned dose due to toxicity, and 1 of those 2 having a prolonged OTT of 55 days. The remaining 5 T1 patients with local failures in the CRT group received 66 Gy in 41—49 days. Median OTT in the CRT group was 46 days. Four patients had OTT > 49 days with one local failure as mentioned above.
Of the 6 failures in the HRT group, all patients completed the full prescribed dose of RT (63 Gy for T1 [n=4] and 65.25 Gy for T2 [n=2]) in 36-42 days. The T2 patient with OTT of 42 days and a CR to HRT failed locally and regionally and was pending surgical salvage at the most recent follow-up. Median OTT in the HRT group was 40 days. Three patients had an OTT >44 days with no local recurrences, and 1 patient received less than the prescribed dose without recurrence.Early toxicity was determined by clinical and laryngoscopic exams performed during RT and at 6-12 week follow-up. Toxicities for all 100 patients were recorded as the maximum toxicity for each endpoint during this time.
Detailed frequency and P-values of patient toxicity by grade is listed in Table 5. CRT patients had less grade 2 dysphagia, grade >2 hoarseness, grade >2 mucositis and required opiate analgesics less frequently. The rates of grade >2 radiation dermatitis, grade >2 laryngeal edema and weight loss were similar.
With the exception of grade 2 dysphagia (P = 0.156), univariate toxicity analysis for the 76 T1 patients confirmed the significance for each specific area noted for the entire cohort. To account for the variable stage distribution of patients between treatment groups, logistic regression of toxicities by UVA and MVA confirmed the significance of these findings (Supplementary Table e1).Nine patients were salvaged surgically including 6 in the CRT group and 3 in the HRT group. There were 6 deaths in the entire cohort: 5 in the CRT group and 1 in the HRT group; none attributable to laryngeal cancer.Although radiotherapy for early glottic cancer is standard treatment for decades, the optimal dose-fractionation scheme to provide maximum local control with minimum treatment toxicity is not clearly established. While several studies have indicated an improvement in local control with HRT compared to lower daily fraction sizes,2-4, 11-13 rarely have these studies directly compared HRT with CRT of 2 Gy fractions. Additionally, most studies have not reported acute treatment toxicity to allow for an accurate assessment of the therapeutic ratio of HRT.
# Fisher’s chi-square test
Numerous retrospective series have demonstrated that local control with 2 Gy fractions (compared to 1.8 Gy fractions) was excellent, ranging from 95-100%.7-9 The University of Florida (UF) experience with hypofractionation using 2.25 Gy fractions developed in the late 1970s and was based on radiobiologic principles without a direct comparison (prospective or retrospective) with CRT using 2 Gy fractions. Mendenhall published the UF retrospective data showing a trend toward improved local control for 4 of 7 small patient cohorts, divided by T stage, who received higher doses per fraction.2 However, 3 of the 4 cohorts with the positive trend compared local control for patients receiving 205-230 cGy per fraction to patients receiving 180-199 cGy per fraction, which had previously been shown to be inferior.7-9, 11 Despite this, the UF experience influenced many other institutions including Yamazaki, who performed the only randomized comparison of 2 vs. 2.25 Gy fractions.
Of note, the only two published retrospective comparisons of CRT to HRT were from Yu12 and Kim.18 The Yu study utilized 2 Gy vs. either 2.25 or 2.5 Gy fractions and showed a local control advantage for hypofractionation. However, the local control rate for the 2 Gy arm was extremely poor at 65.6% at 5 years, approximately 20-30% lower than comparable series. Kim compared 2 vs. 2.25 Gy fractions and showed an advantage for hypofractionation. However, Kim used non-standard doses in the T1 group of 70 Gy for CRT and 67.5 Gy for HRT with prolonged OTT in both arms. In addition, the T2 group in this study showed no advantage to HRT compared to CRT. Importantly, none of the retrospective studies provided detailed analyses of acute treatment toxicity to sufficiently analyze the therapeutic ratio of HRT.
In 2006, Yamazaki et al published a prospective randomized trial of radiation fraction size and OTT, which has since changed practice worldwide by demonstrating a superior local control rate with HRT (2.25 Gy) compared to CRT (2 Gy) with no reported differences in toxicity.14 In this study, 180 patients were randomly allocated to either HRT of 2.25 Gy daily to a total dose of 56.25 Gy for minimal tumors and 63 Gy for larger tumors or to CRT of 2 Gy daily to 60 Gy for minimal tumors and 66 Gy for larger tumors. Minimal tumors were defined as those involving two-thirds of the vocal cord or less. The 5-year local control was significantly improved for HRT (92%) versus CRT (77%) with no significant differences in overall toxicity measured by rates of acute mucosal, skin and chronic adverse reactions, although these toxicities were described in a very limited fashion.
Further criticism of the Yamazaki study includes 5-year local control of only 77% for CRT, well below published retrospective studies, including ours. If the CRT arm had achieved local control results consistent with other studies, there may not have been a significant control advantage with HRT. Regarding toxicity, Yamazaki presents only acute skin and mucosal toxicities, with no mention of hoarseness, dysphagia, pain, or need for narcotics. We found each of these toxicities to be significantly more frequent with HRT. The lack of detailed toxicity information limits analysis of the therapeutic ratio of HRT versus CRT from this single trial.
Several studies analyzing outcome in early glottic cancer by treatment variables have demonstrated that only OTT was significant for local control in MVA, unlike dose and fraction size.3-5,7,11,19,20 Fowler’s review article best summarizes this data and the relevant radiobiological principles.21 In our cohort, the patients receiving CRT consistently completed treatment within the OTT suggested by these studies, which may explain the similar outcomes of CRT to HRT in our study.
Several studies including a recent retrospective series from Japan have indicated that tumor control is inferior with lower total treatment doses.22 Importantly, the combination of overall dose and OTT can significantly influence outcomes. Le et al reported no local failures when the OTT was <50 days and the total dose was >65 Gy but described local failures in 54% of patients when the OTT was >50 days and the total dose was £ 65 Gy (4). A CRT plan of 66 Gy/33 fractions meets these parameters.4
Our outcomes in this series compare favorably with the previously reported rates of local control for conventional fractions of <2 Gy and are similar to outcomes of several studies with a CRT arm of 2-Gy.7-10 Our study found no difference in complete tumor response or local control — described as RFS – rates in a cohort of 100 patients treated over the past decade with either HRT or CRT. This contrasts with the prospective randomized trial from Japan.
Despite the retrospective nature of this study, we have thoroughly analyzed early toxicity by measuring and reporting the maximal severity of toxicity for several key endpoints from the start of radiotherapy through the first follow-up exam at 6 to 12 weeks. Strikingly, our HRT patients developed more grade 2 dysphagia, grade >2 hoarseness, grade >2 laryngeal mucositis, and need for opiate analgesics. These acute toxicities were not routinely recorded or analyzed in the previous retrospective studies or the Japanese randomized study that led to the widespread adoption of HRT.
# Fisher’s chi-square test
Limitations to our series include the retrospective structure and the absence of long-term toxicity data. Patient-reported quality of life data would be desirable, but has not been provided by any investigator in this area or by us. However, our series alone provides detailed information on the severity of acute toxicities of HRT and CRT and thus enables a comparative discussion of the therapeutic ratio of each technique.
We agree with Yamazaki that a multi-institutional prospective randomized trial is essential for HRT to be standard treatment, particularly when toxicity is considered with efficacy. Along with several other investigators, we advocate timely radiotherapy without treatment delay and insist that patients treated with CRT complete their radiotherapy in < 50 days from start of RT.Our retrospective experience suggests that early glottic cancer treated with HRT results in a higher incidence of acute toxicity without improved control compared to CRT delivered in an appropriate OTT. Therefore, HRT may not provide a therapeutic advantage over CRT.
Author affiliations: From the Departments of Radiation Oncology (MTS, NE, MCA, AMM, MAS) and Biostatistics and Bioinformatics Core (WZ), Sylvester Comprehensive Cancer Center at the University of â€¨Miami Miller School of Medicine (AD), Miami, Florida.
Corresponding author: Michael A. Samuels, MD, Department of Radiation Oncology, University of Miami Miller School of Medicine, 1475 NW 12th Avenue, Miami, FL 33136. Tel: (305) 243-4210; Fax: (305) 243-4363; E-mail: firstname.lastname@example.org.
Conflict of Interest Notification: The authors have no conflicts to disclose.