86 / 220
difference in late toxicity
[2]
. This result, combined with data from
the National Cancer Database suggesting a downward trend in
overall survival as chemoradiation was adopted has led some to
consider the possibility that long-term toxicity unaccounted for
on trial led late toxic deaths
[3]
. Late toxicity is difficult to measure
and was reported as a maximum grade in RTOG 91-11 without
comment on the timing or precise nature of the toxic events. The
purpose of this report is to perform a detailed analysis of the
cumulative incidence of late dysphagia observed after concurrent
chemoradiation for larynx cancer in an attempt to identify its con-
tribution to late mortality.
Methods
Patients
From an IRB-approved head and neck database all patients trea-
ted with definitive concurrent chemoradiotherapy at our institu-
tion between 1993 and 2013 who met eligibility criteria for
RTOG 91-11 were retrospectively identified. The criteria included
medically-fit patients with previously untreated squamous cell
carcinoma of the glottis or supraglottis, stage III-IVB by AJCC ver-
sion 3 criteria
[4]
, the version in place during 91-11 accrual.
Patients with disease classified as T4 due to soft-tissue extension
beyond the larynx were excluded from RTOG 91-11 and hence
the current study as well. Patients with T4 disease with
6
1 cm
invasion into the base of tongue were included.
Treatment
All patients were treated with definitive concurrent chemora-
diotherapy. Chemotherapy consisted of multi-agent cisplatin and
5-fluorouracil (5-FU) in the earlier years and single-agent cisplatin
in the later years. Combination chemotherapy was given in two
cycles of a 4-day continuous infusion of cisplatin 20 mg/m
2
/day
and fluorouracil 1000 mg/m
2
/day. Single-agent chemotherapy
was given as a bolus of cisplatin 100 mg/m
2
on weeks 1, 4 and 7
of radiotherapy. All patients were programmed to receive 70 Gy
or greater of radiotherapy. In the early years twice-daily (BID)
radiotherapy was often delivered to doses of 72–74.4 Gy for
tumors of T3 or greater extent along with concurrent cisplatin
and fluorouracil
[5,6]
. Prior to 2009, radiotherapy was delivered
using standard conventional techniques including opposed lateral
fields matched to an anteroposterior supraclavicular field. In
2009, intensity-modulated radiotherapy (IMRT) became the radio-
therapy technique of choice at our institution. After completion of
therapy an adjuvant neck dissection was performed for select
patients within three months for patients with residual adenopa-
thy at the discretion of the treating surgeon. Patients were subse-
quently followed by our multidisciplinary team as per National
Comprehensive Cancer Network (NCCN) guidelines
[7]
.
Severe late dysphagia, other toxicity and the cause of death
The goal of this study was to provide a detailed description of
the incidence and timing of severe late dysphagia (SLD) in this
patient cohort. Therefore, individual multidisciplinary follow-up
encounters documented in the electronic medical record were
investigated for signs of severe dysphagia and other late toxicity.
Ninety days after the completion of radiotherapy was used as the
definition of ‘‘late” events as per RTOG 91-11. Severe late dyspha-
gia was defined as the occurrence of one or more of the following
events 90 days after the completion of treatment or beyond: pha-
ryngeal stricture dilation, admission to the hospital with a diagno-
sis of aspiration pneumonia or placement of a new feeding tube.
Feeding-tube dependence greater than one year after treatment
was also included in the definition of severe late dysphagia. This
composite endpoint is similar to a previous cooperative group
time-to-event analysis of physician-reported severe late dysphagia
[8]
. Furthermore, these events all correlate with the Common Ter-
minology Criteria for Adverse Events (CTCAE) version 4.0 and RTOG
scoring criteria used by RTOG 91-11 but offer a more specific
definition.
In order to evaluate the incidence of death associated with sev-
ere late dysphagia and subsequent aspiration pneumonia, special
attention was paid to the cause of death. Patients who expired after
metastatic or locoregional failure that could not be salvaged were
counted as a primary cancer death regardless of any dysphagia
events occurring after failure. For patients who expired outside of
the hospital system, the death certificate was obtained if possible
through the Ohio death registry.
Statistical analysis
This was a single-cohort retrospective study. Overall survival
and the cumulative incidence of salvage laryngectomy, locore-
gional failure and distant metastatic failure were assessed using
the Kaplan-Meier method measuring time from the date of diagno-
sis to the date of event or last oncologic follow-up. The cumulative
incidence of severe late dysphagia and its four components (eso-
phageal stricture requiring dilation, placement of a new feeding
tube, hospital admission for aspiration pneumonia or feeding tube
dependence beyond one year) was calculated using Gray’s compet-
ing risk analysis treating death or disease recurrence as competing
events
[9]
. Time was measured from the end of radiotherapy to the
first SLD event, with patients censored from the analysis at the
time of the last multidisciplinary follow-up. Although late toxicity
was the primary outcome of interest in this study, patients with
short follow-up or who did not complete therapy were not
excluded in order to capture deaths that may have occurred from
acute toxicity. The cause of death was compared with the data
reported by RTOG 91-11 on a per-patient basis using the Pearson
Chi-square test. To investigate for factors associated with severe
late dysphagia, relevant patient, tumor and treatment variables
were entered into a competing risk regression as described by Fine
and Gray with severe late dysphagia as the endpoint of interest
[10]
. All hypothesis testing was performed with significance
assumed at the 0.05 level. Statistical analyses were performed
using JMP Version 10 software (SAS Institute, Cary, NC) and com-
peting risk analyses were performed using R statistical software
(R Foundation, Vienna, Austria).
Results
Patient characteristics
Between 1993 and 2013, 84 patients were identified who met
the inclusion criteria. Patient, disease and treatment characteristics
are represented in
Table 1
. The majority of tumors were located in
the supraglottis (71%), classified as T3 (78%) and were node-
positive (63%). Combination cisplatin with 5-FU was the most com-
mon concurrent chemotherapy regimen (76%).
The median follow-up for survivors was 54 months (range
8.8–180). Twenty-eight patients (33%) were followed beyond
5 years and 13 (15%) were followed beyond 9 years. A median of
14 individual oncologic follow-up visits were recorded (range
0–44), totaling 1240 individual oncologic visits for the entire
cohort. Fifty patients (60%) have experienced either severe late
dysphagia or a competing event (failure or death) and further
follow-up of these patients will not alter the incidence of severe
M.C. Ward et al. / Oral Oncology 57 (2016) 21–26
66