Treatment Delays in Primarily-Resected Oropharyngeal Squamous Cell Carcinoma: An Analysis of the National Cancer Database

Presentation: AHNS063
Topic: Quality of Care and Clinical Pathways
Type: Oral
Date: Thursday, April 19, 2018
Session: 3:30 PM - 4:30 PM Quality Engineering and Pathways
Authors: Elliot Morse, Shayan Cheraghlou, Benjamin Judson, MD, Saral Mehra, MD, MBA
Institution(s): Yale University Department of Surgery, Division of Otolaryngology

Background: Previous work has suggested that treatment delays are associated with worse overall survival in head and neck cancer (HNC), however oropharyngeal squamous cell carcinoma (OPSCC) is often human papilloma virus (HPV)-associated and behaves differently than other types of HNC. It is unclear if delays are associated with survival.

Objectives: This study characterized treatment delays in OPSCC patients treated with primary surgery, with or without adjuvant radiation. Specifically, this study identified median durations and factors predictive of treatment delays and associated treatment delays with overall survival (OS).

Methods: The National Cancer Database (NCDB) was used to identify a retrospective cohort of OPSCC patients treated 2010-2013 with primary surgery with or without adjuvant treatment. Five treatment intervals were examined: diagnosis-to-treatment initiation (DTI), surgery-to-radiation (SRT), RT duration (RTD), treatment initiation-to-end (total treatment package, TTP), and diagnosis-to-treatment end (DTE). Patients treated with surgery without RT were only included in the analysis examining DTI. For each interval, delayed patients (fourth quartile) were compared to non-delayed patients (first and second quartile). Median durations were identified for each interval. Patient, tumor, and treatment factors predictive of delays were identified via multivariable binary logistic regression. Delays were associated with OS via cox proportional hazards regression, controlling for clinically-relevant patient, tumor, and treatment factors.

Results: 3708 patients were included in the final cohort. Median durations for DTI, SRT, RTD, TTP, and DTE were 14, 42, 47, 90, and 106 days, respectively (Figure 1). Overall HPV incidence was 75%. HPV-positive tumors were associated with decreased delays in DTI, RTD, and DTE (OR=0.69 (0.56-0.85), p<0.001; OR=0.73 (0.58-0.91), p=0.003; OR=0.66 (0.52-0.83), p<0.001, respectively). Overall 5-year OS was 84%. On multivariable analysis, delayed DTI, TTP, and DTE were associated with decreased survival (HR=1.63 (1.22-2.17), p=0.001; HR=1.81 (1.29-2.54), p=0.001; and HR=1.97 (1.39-2.78), p<0.001, respectively) (Table 1; Figure 2). Delayed SRT was associated with survival for HPV-negative, but not positive, patients (HR=2.05 (1.19-3.52), p=0.010 and HR=1.15 (0.74-1.80), p=0.535, respectively). Delays in RTD were not associated with OS.

Conclusions: The median treatment interval durations provided here can be used for national benchmarking. Delays are linked to a variety of patient-, tumor-, and treatment-related factors. Delays in DTI, TTP, and DTE are associated with decreased OS and could be considered quality indicators in OPSCC.

Table 1. Association of Treatment Delays with Survival.
  First and Second Quartiles Fourth Quartile HR (95% CI) p
Days 5-year OS Days 5-year OS
DTI ≤14 87% ≥31 78% 1.63 (1.22-2.17) 0.001
SRT ≤42 88% ≥54 84% 1.31 (0.93-1.85) 0.116
RTD ≤47 89% ≥51 85% 1.40 (0.99-1.99) 0.061
TTP ≤90 89% ≥104 80% 1.81 (1.29-2.54) 0.001
DTE ≤106 90% ≥128 82% 1.97 (1.39-2.78) <0.001

Figure 1. The distribution of treatment delays are shown here. Median durations for each interval are shown to the right of each box. Asterisks depict outliers that were excluded from analyses. DTI=diagnosis-to-treatment initiation. SRT=surgery-to-adjuvant treatment duration. RTD=radiation treatment duration. TTP=total treatment package. DTE=diagnosis-to-treatment end. Figure 2. Association of delays with overall survival. This was analyzed via cox proportional hazards regression, controlling for clincally-relevant patient-, tumor-, and treatment-related characteristics. Delayed patients (fourth quartile) were compared to non-delayed patients (first or second quartiles) for each interval. p-values for the effect of delay are shown to the right of each chart.