Modern Image-Guided Intensity-Modulated Radiotherapy for Oropharynx Cancer and Severe Late Toxic Effects: Implications for Clinical Trial Design

Key Points

Question  What is the incidence of severe late toxic effects after modern definitive image-guided intensity-modulated radiotherapy (IG-IMRT) in the era of human papillomavirus-associated oropharynx cancer?

Findings  In this review of 156 patients with stage I-IVB squamous cell carcinoma of the oropharynx, the cumulative incidence of severe late dysphagia after narrow-margin IG-IMRT for oropharynx cancer is reported to be very low at 2 years.

Meaning  This finding has implications for informed consent discussion, patient selection for deintensification protocols, and for future clinical trial design.

Importance  Late toxic effects are common after definitive radiotherapy and chemoradiotherapy for oropharynx cancer and are considered a significant contributor to decreased quality of life for survivors. The incidence of severe late toxic effects may be reduced by modern narrow-margin image-guided intensity-modulated radiotherapy (IG-IMRT), current supportive care improvements, and the changing epidemiology of oropharynx cancer.

Objective  Assess the incidence of severe late toxic effects after modern definitive non-operative treatment for oropharynx cancer.

Design, Setting, and Participants  For this single-institution retrospective review, 156 patients with stage I-IVB squamous cell carcinoma of the oropharynx treated between April 2009 and February 2015 at a tertiary-referral academic multidisciplinary head and neck practice were recruited.

Interventions  Definitive narrow-margin IG-IMRT to a dose of 66 Gy (to convert milligray to rad, multiply by 0.1) or higher with or without concurrent cisplatin.

Main Outcomes and Measures  The primary outcome was the prospectively collected 2-year cumulative incidence of severe late toxic effects (Common Terminology Criteria for Adverse Events grade 3 or higher) occurring 3 months or more after radiotherapy. Toxic effect end points investigated included esophageal stricture requiring dilation, aspiration pneumonia hospitalization, vocal dysfunction, delayed feeding tube insertions, and osteoradionecrosis. Feeding tube dependence at 1 year was also considered a severe late toxic effect. Secondary outcomes collected include physician-reported grade 2 or higher neck fibrosis and xerostomia. The competing risks of recurrence and death were accounted for using the Gray method.

Results  One-hundred fifty-six patients (median [range] age, 58 [37-96] years) were identified; 130 patients (83%) were HPV positive. Concurrent cisplatin was delivered in 131 patients (84%) and 5 patients (3%) underwent an adjuvant neck dissection. The median (range) follow-up for survivors was 22 (4-73) months from diagnosis. The projected 2-year locoregional control was 93% (95% CI, 88.4%-97.6%) and overall survival was 88% (95% CI, 82.2%-94.0%). Thirty-eight patients (23%) required a feeding tube during treatment. The cumulative incidence of severe late toxic effects adjusted for competing risks at 2-year posttreatment was 2.3% (95% CI, 0%-5.6%). One patient required free-flap reconstruction for grade 3 osteoradionecrosis at 47 months. At 1 year, 2 patients (1%) experienced grade 2 neck fibrosis and 38 patients (23%) experienced grade 2 xerostomia.

Conclusions and Relevance  These results suggest that severe late toxic effects after modern definitive IG-IMRT, with or without cisplatin, for oropharynx cancer is likely uncommon. The importance of late toxic effect reduction in current and future investigational strategies, including clinical trials, should be considered.

Given the morbidity of classic open surgical techniques, radiotherapy with or without concurrent cisplatin-based chemotherapy has become a standard option for the treatment for oropharynx cancer. Conventional radiotherapy and chemotherapy, however, results in significant late toxic effects in up to 43% of patients.^1-3 Late toxic effects such as dysphagia, radionecrosis, and xerostomia can have a significant and often permanent effect on long-term quality of life.⁴

Mainly owing to the rapidly increasing incidence of good-prognosis human papillomavirus (HPV)-induced oropharynx cancer, long-term survival rates for oropharyngeal cancer have dramatically improved. Nonsmoking patients with newly diagnosed locoregionally advanced HPV-induced oropharynx cancer can now anticipate cure rates in excess of 90% after chemoradiotherapy, emphasizing the concern regarding late toxic effects.^5,6 As a result, multiple clinical trials⁷ (NCT02254278) have been initiated evaluating strategies such as transoral robotic surgery and chemoradiotherapy deintensification in an effort to reduce late toxic effects without compromising survival in these favorable-risk patients.

Compared with conventional radiation techniques, intensity-modulated radiotherapy (IMRT) delivers a better dose distribution to the target volume while limiting dose to the nearby critical structures, thereby reducing overall toxic effects.^8,9 The technique has continued to evolve over recent years owing to improved planning algorithms and the reduction of treatment margins allowed by daily image guidance (IG).¹⁰ These improvements, coupled with better supportive care and the changing disease epidemiology would be expected to even further reduce the incidence of severe late toxic effects beyond the initial IMRT reports. This study provides a detailed analysis of severe late toxic effects after definitive radiotherapy with or without chemotherapy for oropharynx cancer using modern narrow-margin IG-IMRT.

Patients with stage I to IVB squamous cell carcinoma of the oropharynx treated between April 2009 to February 2015 with definitive radiotherapy with or without concurrent single-agent cisplatin were identified from a local institutional review board (IRB)-approved head and neck cancer registry of Cleveland Clinic patients. The Cleveland Clinic IRB provided approval for a protocol that received a waiver for informed consent. Oropharyngeal cancer was defined as originating from the base of tongue, tonsil, soft palate, and posterior oropharyngeal wall. Patients with pretreatment tracheostomy or feeding tube dependence were excluded from this analysis; prophylactic feeding tubes are not used within the Cleveland Clinic. Comorbidity was graded according to the Charlson comorbidity index.¹¹ Human papillomavirus status was determined in the early study period by simultaneous fluorescence in situ hybridization (FISH) for HPV DNA and immunohistochemical analysis for the p16 protein. In the later years (approximately 2012-2015) HPV status was determined by p16 status alone. Patients treated with noncisplatin chemotherapy regimens or non-IMRT techniques were excluded.

Patients were treated with definitive IMRT to doses of 66 Gy (to convert milligray to rad, multiply by 0.1) or higher with or without concurrent cisplatin. In general, cisplatin was given to healthy patients with American Joint Committee on Cancer stage III to IVB disease at a dose of 100 mg/m² every 3 weeks. A minority of patients received weekly cisplatin doses of 40 mg/m². Patients treated with the monoclonal antibody cetuximab were not included. Quiz Ref IDIntensity-modulated radiotherapy planning was performed using a gross tumor volume (GTV) to clinical target volume (CTV) expansion of 2 to 3 mm. This expansion was then trimmed out of air, muscle, and bone and an additional CTV to planning target volume (PTV) expansion of 2 to 3 mm was added for a total GTV to PTV expansion of 5 to 6 mm.Quiz Ref ID The elective neck was treated in all patients, including those with stage I to II disease. Unilateral neck irradiation was applied for tonsillar carcinomas with a well-lateralized primary tumor as defined by O’Sullivan et al.¹² The typical dose to the elective uninvolved neck was 56 Gy in 35 fractions at 1.6 Gy per fraction. Daily image guidance was used in all cases—with megavoltage cone-beam computed tomography in the early years of the study and with kilovoltage cone-beam computed tomography in the later years. All patients were treated with 6 megavoltage photons with either static IMRT fields or with volumetric arc therapy delivered using 5-mm multileaf collimator leaves. A second simulation and treatment plan (adaptive replan) was allowed when necessary at physician discretion but was generally performed when neck adenopathy exceeded 3 cm in diameter.

Oncologic outcomes investigated include overall survival (OS), locoregional control (LRC) and distant metastases rate. Acute and late toxic effects were graded prospectively according to the Common Terminology Criteria for Adverse Events (CTCAE) v4.0 criteria. Acute toxic effects were considered those events occurring within 90 days after of the completion of radiotherapy. Acute events specifically collected included dysphagia, feeding tube use, aspiration, tracheostomy use, and neutropenic fever.

Quiz Ref IDThe primary end point of the study were severe late toxic effects occurring 90 days or more after the completion of radiotherapy. Severe late toxic effects were defined as CTCAE grade III or higher, including esophageal stricture requiring dilation, hospital admission for aspiration pneumonia, delayed feeding tube placement more than 90 days after treatment completion, voice dysfunction (tracheostomy use or a whispered voice unable to be heard on the phone), or osteoradionecrosis requiring hyperbaric oxygen or surgical intervention. Feeding tube dependence beyond 1 year was also considered a severe late toxic effect. Secondary outcomes included prospectively collected, physician-reported CTCAE grade II or higher xerostomia and neck fibrosis if the symptoms were present 1 year after treatment or beyond.

The Kaplan-Meier technique was used to calculate OS, locoregional control, and distant metastases rate. All survival outcomes were calculated from the date of diagnosis until event occurrence and were censored at the date of last contact. The event for OS was death from any cause. Events for locoregional control were primary site or cervical node recurrence. The event for distant metastases was the development of any disease recurrence beyond the cervical nodes. The Gray competing risk model¹³ was used to calculate the cumulative incidence of severe late toxic effects, counting recurrence and death as competing events. For the calculation of late toxic effects, patients were censored at the last date of clinical follow-up with a head and neck cancer specialist. All statistical calculations were performed using JMP Pro version 10 (SAS institute) and competing risk analyses were performed using R version 3.2.3 (R foundation).

Between April 2009 to February 2015, 156 patients who met the inclusion criteria were treated. Table 1 presents the demographic and disease factors. The median (range) age was 58 (37-96) years, 132 patients (84%) were male, and 139 (89%) patients were white. Performance status was generally excellent and most patients were nonsmokers. Figure 1 presents the clinical T (size of primary tumor and whether it has invaded nearby tissue) and N (regional lymph nodes involved) American Joint Committee on Cancer classification of the disease. The base of the tongue was the most common primary tumor site in 77 patients (49%); 145 patients (93%) had stage III to IVB disease, and 130 patients (83%) were HPV positive.

No patients included were initially dependent on a feeding tube or tracheostomy. Twenty-eight patients (18%) were experiencing symptomatic dysphagia not requiring a feeding tube at presentation although their diet had been altered to require soft foods or protein supplements at baseline (CTCAE grade 2 baseline dysphagia). Nineteen patients (12%) were experiencing voice hoarseness at baseline that was persistent but did not require a tracheostomy and was audible over the telephone (CTCAE grade 2 baseline hoarseness).

Table 2 details the treatment received by the study population. The median (range) prescription dose of radiotherapy was 70 (66-78) Gy. The median (range) duration of treatment was 45.5 (38-63) days. The majority of patients (n = 140 [90%]) received bilateral neck irradiation while 16 patients (10%) received unilateral neck irradiation. Fractions were administered once daily for a total of 5 fractions per week (QD) in 88 patients (56.4%), daily with a second fraction given once a week for a total of 6 fractions per week (accelerated QD¹⁴) in 67 patients (42.9%), and 1 patient received 2 fractions per day (BID). Adaptive replanning was utilized in 65 patients (42%). Concurrent cisplatin was delivered in 131 patients (84%) and 25 patients (16%) were treated with radiation alone. Cisplatin was given using a high dose every 3-week schedule in 119 patients (91%), and was given weekly in 12 patients (9%). Only 5 patients underwent adjuvant neck dissection after radiation. The median (range) time from the end of therapy to neck dissection was 4.97 (4.38-5.43) months.

The median (range) length of follow-up was 22.3 (4.0-73.0) months in surviving patients. Seventy-six patients (49%) had a total follow-up longer than 2 years or experienced a competing risk event of failure or death prior to 2 years. The 2-year OS rate was 87.9% (95% CI, 82.2%-94.0%); locoregional control rate, 92.9% (95% CI, 88.4%-97.6%); and distant metastasis rate, 11.9% (95% CI, 81.8%-93.8%). Eleven patients died of head and neck cancer, 2 died of other cancers, and 4 died for unknown reasons.

Rates of acute toxic effects were moderate and consistent with other reports in the literature.^15-17 Overall, 66 of 156 patients (42%) experienced at least 1 grade 3 or higher acute toxic effect. Quiz Ref IDThe most common grade 3 or higher toxic effect was dysphagia and was observed in 37 patients (24%). Twenty-six patients (17%) had grade 3 mucositis, and 12 patients (8%) had grade 3 dermatitis. Grade 2 or higher acute dysphagia was observed in 89 patients (55%) of patients, and 36 patients (23%) required a feeding tube during treatment or within the first 90 days after treatment. Nasogastric tubes were used for 75% of those who required a feeding tube and gastrostomy tubes in the remaining 25%. A tracheostomy was required during treatment in 4 patients (3%). No patient was admitted to the hospital for aspiration pneumonia during treatment. Ten patients (6%) experienced neutropenic fever requiring hospitalization.

Quiz Ref IDOverall, severe late toxic effects were uncommon. A total of 3 patients experienced severe late toxic effects for a 2-year cumulative incidence of 2.3% (95% CI, 0.0%-5.6%).Figure 2 presents the cumulative incidence of severe late toxic effects. One patient with a T3N0 HPV-negative carcinoma of the posterior oropharyngeal wall developed a nonhealing, biopsy specimen–proven, noncancerous ulceration of the posterior oropharyngeal wall that required delayed feeding tube placement at 6 months postradiation, with subsequent surgical repair and esophageal dilations. The second patient developed dysphagia 22 months postradiation requiring temporary gastrostomy tube placement, hospital admission for aspiration pneumonia, and esophageal dilation. The third patient developed mandibular osteoradionecrosis 4 years after treatment. No other severe late toxic effects occurred.

The 3 patients who developed toxic effects were 68, 61, and 58 years old, respectively. None of the 19 elderly patients (70 years or older) developed severe late toxic effects. Although all patients experiencing toxic effects were at or above the median age of the cohort, no clear statistically significant effect of age could be determined given the small number of patients experiencing severe late toxic effects.

Figure 3 demonstrates the duration of acute feeding tube use for those feeding tubes placed during treatment. The median duration of acute feeding tube use was 1.7 months after the end of radiotherapy; at 6 months after radiotherapy, all but 2 feeding tubes placed during therapy were removed (6-month feeding-tube dependence rate, 2.0%; 95% CI, 0.7%-6.0%). All tubes placed during treatment had been removed within the first year after treatment. Of patients with follow-up beyond 1 year who maintained locoregional control, 37 of 94 (24%) were experiencing grade 2 xerostomia and 2 of 94 patients (1%) reported grade 2 neck fibrosis. The low rate of severe late toxic effects precluded further univariate or multivariate regression analyses. A clear contribution of chemotherapy to severe late toxic effects could not be identified.

This study reports the late toxic effects after modern definitive nonoperative treatment of oropharynx cancer and demonstrates that disease control is excellent, and severe late toxic effects are markedly reduced compared with historical reports. Future protocols seeking to reduce late toxic effects by deintensifying chemotherapy or radiation treatment or by reintroducing transoral surgery should consider these results, as full-dose radiotherapy with or without chemotherapy appears to lead to acceptable rates of severe late toxic effects in the modern era.

The benchmark data identifying the incidence of severe late toxic effects after chemoradiotherapy was reported by Machtay et al,¹⁵ who analyzed 3 Radiation Therapy Oncology Group (RTOG) trials using conventional radiation and concurrent chemotherapy. The severe late toxic effects rate, defined similarly to the current study, was 43%.¹⁵ A more recent trial, RTOG 0129,¹⁶ performed in the conventional radiotherapy era, demonstrated a grade 3 or higher late toxic effect rate of approximately 25%.¹⁶ Another trial, RTOG 0522,¹⁷ performed in the early IMRT era, demonstrated a rate of feeding tube dependence of approximately 20% at 1 year.¹⁷ Although a direct comparison of the difference in late toxic effect rate between studies should be approached with caution owing to variance in inclusion criteria and measurement of toxic effects, it is clear that the rates of severe late toxic effects we observed are much less than these historical outcomes.¹⁸

The use of modern narrow-margin IG-IMRT may be a key component in the reduction of severe late toxic effects for these patients. Initial prospective evaluation of IMRT suggested the need for an expansion from the GTV to CTV of 1 to 2 cm with an additional CTV to PTV expansion of 5 mm.¹⁹ In comparison, the current study applies a total GTV to PTV expansion of only 5 to 6 mm. This has been shown to be both dosimetrically and clinically safe when treating with daily image-guidance to ensure minimal setup uncertainty.^20-22 We suggest that for these patients the volume of tissue irradiated may be as important in the reduction of late toxic effects as the total dose delivered to the tumor.

Other factors besides IMRT technology have likely also contributed to this reduction in severe late toxic effects, including the decrease in tobacco use,²³ the increased incidence of HPV-positive disease,^24,25 and improved supportive care. The demographics of contemporary patients with oropharynx cancer have shifted. Patients are now younger, HPV-positive, with limited or no smoking history, and with less tobacco related comorbidity.^6,26-28 Smoking has previously been shown to increase the rate of late toxic effects, and a decreasing prevalence of tobacco abuse may account for some of the reduction in late toxic effects seen in our cohort.²⁹ Similarly, the biologic differences between HPV-positive and HPV-negative oropharynx cancer may also independently affect late toxic effects.^30,31 The increase and improvement in proactive supportive care measures may also have decreased late toxic effect rates independent of technology.³²

Several ongoing clinical trials have been initiated with the goal of decreasing the rate of severe late toxic effects without compromising the excellent survival outcomes currently expected. For example, one trial (NCT01893307) comparing IMRT to proton therapy seeks to reduce the 2-year cumulative incidence of grade 3 or higher late toxic effects as a primary end point. Another study (NCT01898494) investigating the use of transoral robotic surgery includes adverse events within 3-year as a secondary end point. A third study (NRG HN-002) is investigating the safety of dose-reduced IMRT with or without chemotherapy and includes late toxic effects as a secondary end point. Given the very low rates of severe late toxic effects observed in this present study, we raise concern that studies investigating physician-reported late toxic effects may be underpowered and that a more sensitive end point may be required.

There are several limitations to this study. Additional severe late toxic effects are likely to be observed with increased follow-up, particularly in these good-prognosis HPV-positive patients with a high likelihood of cure. This study also has the characteristic limitations of all retrospective studies, including the potential for selection and recall bias. Over time there is a certain inhomogeneity in the assessment of dysphagia, making it somewhat problematic to compare long-term dysphagia rates longitudinally and against other studies. It is also possible that preexisting subclinical dysphagia was not detected on initial physician-reported assessments. Although efforts to standardize our assessment of dysphagia were made, prospective collection of patient-reported outcomes would prove valuable in future studies.

These results suggest that the risk of severe late toxic effects is low within the first 2 years after treatment in the modern nonoperative management of oropharynx cancer. Future clinical trials seeking to reduce severe late toxic effects should consider this decreased risk in the trial design. The importance of late toxic effect reduction in current investigational strategies may be overemphasized, but additional follow-up after modern IG-IMRT is warranted.

Corresponding Author: Matthew C. Ward, MD, Department of Radiation Oncology, Taussig Cancer Institute, Cleveland Clinic Foundation, 9500 Euclid Ave, Desk T28, Cleveland, OH 44195 (wardm3@ccf.org).

Published Online: July 20, 2016. doi:10.1001/jamaoto.2016.1876

Author Contributions: Dr Ward had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Ward, Koyfman, Lamarre, Joshi, Greskovich, Adelstein.

Acquisition, analysis, or interpretation of data: Ward, Ross, Koyfman, Lorenz, Scharpf, Burkey, Woody, Prendes, Houston, Reddy, Greskovich, Adelstein.

Drafting of the manuscript: Ward, Ross, Adelstein.

Critical revision of the manuscript for important intellectual content: Ward, Ross, Koyfman, Lorenz, Lamarre, Scharpf, Burkey, Joshi, Woody, Prendes, Houston, Reddy, Greskovich, Adelstein.

Statistical analysis: Ward, Reddy.

Administrative, technical, or material support: Lorenz, Scharpf, Burkey, Houston.

Study supervision: Ward, Lamarre, Scharpf, Woody, Adelstein.

Discussion of Complications and Implications: Burkey.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

Previous Presentation: This study was presented at the American Head & Neck Society Ninth International Conference on Head and Neck Cancer; July 20, 2016; Seattle, Washington.