We retrospectively reviewed the medical records of patients who were diagnosed with early stage squamous cell carcinoma of the glottic larynx between January 2006 and December 2012 in two major clinical institutes. Early stage glottic cancer was defined as stage Tis (in situ), T1-2 without nodal metastases. Patients diagnosed with double primary cancer (n=35), and patients who received treatment prior to RT or cordectomy (n=194) or received both treatments (n=39) were excluded. The remaining 165 patients were retrospectively analyzed.

All patients were diagnosed by laryngeal microsurgery biopsy and clinically staged according to the American Joint Committee on Cancer seventh edition cancer staging.

Patients who were considered not adequately resectable by cordectomy were referred for RT. Of the 165 patients, a total of 112 received definitive RT. The median dose for definitive RT was 65.25 Gy (range, 58.5 to 69.75 Gy). The most commonly used dose scheme was 64 Gy in 32 fractions for T1 stage and 66 Gy in 33 fractions for T2 stage glottic cancer patients. The fractional dose was 2 Gy (n=54), 2.25 Gy (n=56), or other (n=2). Two patients grouped as other received ipsilateral vocal cord irradiation with a fractional dose of 1.8 Gy or 2.3 Gy via tomotherapy. Definitive RT was performed by conformal RT (n=105) or tomotherapy (n=7). The gross tumor was delineated as gross tumor volume (GTV). The clinical target volume (CTV) was delineated including the whole larynx. Planning target volume (PTV) was defined as the CTV plus an adequate margin to encompass the possible setting uncertainties. For patients who received conformal RT, a traditional method was used; namely, a box field with the superior border at the top of the thyroid cartilage, the inferior border at the bottom of the cricoid cartilage (for T2, one tracheal ring below the cricoid cartilage), the anterior border at a 1 cm skin flash anteriorly, and the posterior border at the anterior edge of the vertebral body. No cone down was performed in 22 patients, and the RT field was reduced to the ipsilateral vocal cord in 20 patients. For the other 63 patients, the box field size was reduced by 1 to 2 cm depending on the patient. The median cone down dose was 46 Gy (range, 36 to 54 Gy). For those who received tomotherapy, but not ipsilateral vocal cord irradiation, the initial field was the delineated PTV, after which cone down was performed to GTV plus margin after 56 Gy.

A total of 53 patients received cordectomy, which was performed by one of two modalities: transoral laser excision (n=44) or transoral robotic surgery (n=9). Cordectomy types were type I (subepithelial cordectomy, n=9), type II (subligamental cordectomy, n=10), type III (transmulscular cordectomy, n=5), type IV (total cordectomy, n=1), type V (extended cordectomy, n=1), and unknown (n=27). The type of cordectomy was determined by the physician. No patients received chemotherapy during or after treatment in either the RT or the cordectomy group.

Follow-up was performed regularly in both the Departments of ENT and Radiation Oncology on an outpatient clinic basis. The follow-up schedule was 1 month after completion of treatment and every 3 months for the first year, every 6 months for 5 years, and then annually thereafter. Events of recurrence, metastasis, and toxicity were recorded during follow-up visits. All outpatient clinic reports were retrospectively reviewed in detail to identify any toxicity events.

The primary endpoint was LC, and secondary endpoints included DFS, OS, and larynx preservation rates. Larynx preservation was defined as preservation of the larynx without undergoing total laryngectomy. Laryngectomy types for larynx preservation such as supracricoid laryngectomy or vertical partial laryngectomy were not considered failures of larynx preservation. The cumulative probability of LC, DFS, OS, and larynx preservation rates were calculated by the Kaplan-Meier method and compared by the log-rank test. Univariate and multivariate analyses of LC were performed using Cox regression. Multivariate analysis was performed using a backward elimination approach including all of the variables.

Patient and tumor characteristics are shown in Table 1. The median age was the same for both groups, and patient characteristics such as age, sex, tumor type, histologic grade, anterior commissure involvement, location, and smoking history were well balanced between groups (all p > 0.05). There was a significant difference between the definitive RT and cordectomy group in T stage (p=0.001). In the RT group, 90 patients (80.4%) had T1/is and 22 patients (19.6%) had T2 stage disease. However, all 53 patients (100.0%) in the cordectomy group had T1/is. There were three stage Tis patients in the RT group and eight in the cordectomy group. Tumor type also differed significantly between groups (p=0.022). While the dominant tumor type in the RT group was the exophytic type, the dominant type in the cordectomy group was the superficial type. For the cordectomy group, a total of 26 out of 53 cordectomy cases reported the status of resection margin as positive or negative, and most did not report the length of the safety margin.

The median follow up period was 77.7 months (range, 10.7 to 127.0 months). LC was the primary endpoint. The 3- and 5-year LC rates were 91.9% and 89.9% for the RT group and 82.8% and 73.2% for the cordectomy group, respectively (p=0.006) (Fig. 1A). Since no T2 patients received cordectomy, a subgroup analysis of only T1 glottic cancer patients was conducted. For T1 patients (n=143), the 3- and 5-year LC rates were 95.6% and 93.0%, respectively, for the RT group (n=90), and 82.8% and 73.2% for the cordectomy group (n=53) (p=0.001) (Fig. 1B). Multivariate analysis revealed that LC was adversely affected by a T stage of T2 (hazard ratio [HR], 5.36; 95% confidence interval [CI], 1.44 to 19.99; p=0.012) and cordectomy as a treatment modality (HR, 7.79; 95% CI, 2.35 to 25.87; p=0.001) (Table 2).

The 3- and 5-year DFS rates were 87.5% and 83.7%, respectively, for the RT group and 79.2% and 68.0%, respectively, for the cordectomy group (p=0.046) (Fig. 2). A subgroup analysis of T1 glottic cancer patients showed similar results with 3- and 5-year DFS rates of 91.9% and 87.0%, respectively, for the RT group and 75.5% and 64.8%, respectively, for the cordectomy group (p=0.007). For OS, no significant differences were identified between the 3- and 5-year OS. The 3- and 5-year OS rates were 94.6% and 92.8%, respectively, for the RT group and 94.3% and 90.6%, respectively, for the cordectomy group (p=0.713) (Fig. 3).

Two patients in the RT group and one in the cordectomy group died of disease progression. Other causes of death were a newly developed second malignancy (n=8), postoperative bleeding after salvage surgery (n=1), aspiration pneumonia (n=1), and unknown because of follow-up loss with no evidence of disease (NED) status (n=11). The newly developed second malignancies were hepatocellular carcinoma, cholangiocarcinoma, pancreatic cancer, esophageal cancer, and lung cancer.

In the definitive RT group, a total of 12 local failures, three regional failures, and one distant failure occurred. All failures occurred independently. For local failures, failure sites included the involved vocal cord. For patients with recurrent stage T1 or T2N0M0, salvage treatment modalities were cordectomy (n=4), re-irradiation (n=1), supracricoid partial laryngectomy (n=1), and none because of a second primary cancer (n=1). Among four patients who received salvage cordectomy, two were salvaged successfully, one died of lung cancer, and one died due to disease progression. For recurrent stage T3 or T4N0M0, salvage treatment modalities were total laryngectomy (n=3), supracricoid partial laryngectomy (n=1), and none because of follow-up loss (n=1). Among regional failures, two were treated with salvage neck dissection followed by postoperative RT and one was treated with salvage chemoradiotherapy. One distant failure that occurred in the T-spine was successfully salvaged by RT.

In the cordectomy group, a total of 14 local failures and one distant failure occurred. Almost all local failure sites included the ipsilateral vocal cord, except for one failure that occurred in the contralateral vocal cord. One distant failure occurred as a hilar lymph node metastasis and was treated with RT. Salvage treatment modalities performed for local recurrences were laryngeal microsurgery (n=2), vertical partial laryngectomy or supracricoid partial laryngectomy (n=3), cordectomy (n=3), RT (n=5), and none because of follow-up loss (n=1). All five patients who received salvage RT were salvaged successfully.

Larynx preservation rates showed no significant difference between the groups. For the RT group, the 3- and 5-year larynx preservation rates were 98.2%, while they were 100.0% and 97.2% in the cordectomy group (p=0.831).

During RT, three patients experienced a grade III skin reaction and two experienced grade III pharyngitis. Additionally, one patient required a 1-week break during treatment because of a grade III skin reaction. There were no late complications more severe than grade 2. When follow-up of the patients in the RT group was conducted in the radiation oncology department, subjective scoring of the voice quality was performed. The patient scored the voice quality in a scale of 0 to 100. The score for most patients was more than 80, but six patients (5.4%) still complained of continued hoarseness at a median of 10 months (range, 3 to 44 months) after treatment. In the cordectomy group, in one institution, patients who complained of severe voice impairment underwent assessment that included speech intelligibility and the percentage of consonants correct as criteria, which resulted in three patients being diagnosed with phonation disorder. In the other institution, voice handicap index was assessed for two-thirds of the patients who received cordectomy, and a total of six patients (11.3%) complained of moderate to severe voice impairment at a median of 13 months (range, 2 to 46 months) after cordectomy.