The natural history of patients with squamous cell carcinoma of the hypopharynx.

Abstract

To provide the baseline information on the natural history of patients with squamous cell carcinoma of the hypopharynx to help clinicians, researchers, and patients assess the relative effectiveness of treatment options when the best treatment is not known and newer treatments are being proposed.
Retrospective population-based design.
The patient descriptors, treatments, and outcomes for 595 patients across the province of Ontario, Canada from January 1990 to December 31, 1999 based on electronic data and chart review.
The typical patient is 65 years old, male, unemployed, and poor. They are heavy drinkers with significant comorbidity compromising functional status. The tumors are advanced (over 50% stage 4). After curative treatment 20% had residual disease, recurrences tended to appear in the first year and 50% of first recurrences included metastases. Overall, 47% of patients were disease free at 3 years but eventually 64% of patients died of their cancer.
This information can be used by clinicians and researchers to understand the natural history of the patient group to critically assess both the selection bias and effectiveness of treatments.

Full text

Page 1

The Laryngoscope
Lippincott Williams & Wilkins
© 2008 The American Laryngological,
Rhinological and Otological Society, Inc.
The Natural History of Patients
With Squamous Cell Carcinoma of
the Hypopharynx
Stephen F. Hall, MSc, MD, FRCSC; Patti A. Groome, PhD; Jonathan Irish, MSc, MD, FRCSC;
Brian O’Sullivan, MB, FRCPC
Objectives/Hypothesis: To provide the baseline
information on the natural history of patients with
squamous cell carcinoma of the hypopharynx to help
clinicians, researchers, and patients assess the rela-
tive effectiveness of treatment options when the best
treatment is not known and newer treatments are
being proposed.
Study Design:
Retrospective
design.
Methods: The patient descriptors, treatments, and
outcomes for 595 patients across the province of Ontario,
Canada from January 1990 to December 31, 1999 based
on electronic data and chart review.
Results: The typical patient is 65 years old, male,
unemployed, and poor. They are heavy drinkers with
significant comorbidity compromising functional status.
The tumors are advanced (over 50% stage 4). After cura-
tive treatment 20% had residual disease, recurrences
tended to appear in the first year and 50% of first recur-
rences included metastases. Overall, 47% of patients
were disease free at 3 years but eventually 64% of pa-
tients died of their cancer.
Conclusions: This information can be used by cli-
nicians and researchers to understand the natural
population-based
history of the patient group to critically assess both the
selection bias and effectiveness of treatments.
Key Words: Squamous carcinoma, hypopharynx,
natural history, population-based research.
Laryngoscope, 118:1362–1371, 2008
INTRODUCTION
In planning the treatment of a patient with squa-
mous carcinoma of the hypopharynx, head and neck on-
cologists are faced with three significant obstacles. First,
the patients are complicated, often with advanced stage
disease combined with both debilitation secondary to the
cancer and associated significant comorbid illnesses. Sec-
ond, there is no level one evidence on best treatment1or
agreement on treatment2,3and finally there is a lack of
understanding of the natural history of this terrible can-
cer. Our lack of understanding is due to the low incidence,
the poor prognosis of this patient group, and the outcomes
biased by the treatment selection of observational studies.
Hypopharynx cancer is not common. It accounts for 5% of
all head and neck cancers that themselves account for 5%
of all cancers so that across the province of Ontario, Can-
ada (13 million people), there are less than 100 new cases
per year. With a small number of cases, an individual
oncologist, even in a busy center, has a limited experience
over a lifetime diluted by the more common cancers of the
larynx, oropharynx, and oral cavity. Of the cases seen and
treated, many have multispecialty treatments and as
most seem to do poorly they are quickly lost to follow-up by
the initial treating physicians. This study was designed to
improve the understanding of the natural history by de-
scribing the patients, the tumors, and the disease course
based on a large complete population of patients.
The objective of this study is to provide the baseline
information on the natural history of patients to help
clinicians, researchers, and patients assess the relative
effectiveness of treatment options. The clinical and re-
search environment for this study was the evolution of
From the Departments of Otolaryngology and Oncology (S.F.H.),
The Division of Cancer Care and Epidemiology, Queen’s Cancer Re-
search Institute (P.A.G.), Queen’s University, Kingston, Ontario, Can-
ada; and Departments of Otolaryngology and Surgical Oncology (J.I.),
and Radiation Oncology (B.O.S.), Princess Margaret Hospital, Toronto,
Canada.
Editor’s Note: This Manuscript was accepted for publication March
13, 2008.
Supported by Canadian Institutes of Health Research (CIHR)
MOP 62781.
Dr. Hall is a CIHR New Investigator; Dr. Groome is the Canada
Research Chair in Cancer Care Evaluation; Tina Dyer, CHIM, CTR, was
the study coordinator; and Yan Ouyang, MSc, was the statistician.
Send correspondence to Stephen F. Hall, MSc, MD, FRCSC, 10
Stuart St, Level 2, Kingston, ON K7L 3N6, Canada. E-mail: sfh@
post.queensu.ca
DOI: 10.1097/MLG.0b013e318173dc4a
Laryngoscope 118: August 2008
1362
Hall et al.: Squamous Cell Carcinoma of the Hypopharynx

Page 2

and the nonagreement on the best treatment for these
unfortunate patients. The results of both the traditional
treatment options of surgery or radiotherapy (RT) have
improved with better assessment of the extent of disease,
improved surgical reconstruction, and improved RT
technologies. Planned preoperative RT evolved to the
very common treatment of planned postoperative RT
and was then followed by advances in chemotherapy
initially as induction chemotherapy. The present treat-
ments options include regimes incorporating concomi-
tant chemotherapy with RT, intensity-modulated radi-
ation therapy and postoperative chemotherapy with RT,
but the relative effectiveness of these treatments is
difficult to assess when case numbers are small, case
series are biased by selection and clinical trials are few.
The only reported randomized trial in hypopharynx cancer
compared induction chemotherapy with RT (for responders)
to surgery with postoperative traditional RT.4The trial was
designed as an equivalence trial and found no statistically
significant difference between the treatments for survival or
recurrence. There was no RT arm and more patients re-
tained a functioning larynx in the chemotherapy arm. The
importance of this trial was the evidence on organ preserva-
tion and although the protocol became the standard of treat-
ment for European Organization for Research and Treat-
ment of Cancer, North America, has continued to search for
better drugs, protocols, fractionations, and combinations of
treatments.
We have collected and analyzed the patient descrip-
tors, treatments, and outcomes for 595 patients across the
province of Ontario Canada from January 1990 to Janu-
ary 2000. We chose this timeframe for two specific rea-
sons. First, treatment policies were stable as concomitant
chemotherapy with RT had not been adopted in Ontario,
and second, stage migration due to evolving imaging tech-
nology (magnetic resonance imaging) would be minimal as
computed tomography (CT) scan was the imaging stan-
dard for all centers.
For the treatment of all patients with cancer, the
province of Ontario is divided into eight geographic re-
gions each with at least one regional treatment center that
is commonly associated with a teaching hospital. The
overall treatment philosophy and patient treatment selec-
tion varied widely from center to center and to avoid the
center-specific selection bias; all patients with all treat-
ments from all the regional cancer centers are combined
for this study. Overall the initial treatments were sur-
gery ? RT (19.7%), RT ? chemotherapy ? surgical sal-
vage for residual disease (64.6), palliative RT (5.9%), and
no active treatment (9.7%). The range of treatments be-
tween centers was 41.7% to 82.2% for RT and 4.1% to
41.8% for surgery.
There are only three large studies with unselected
patients in the literature that can be used for comparison
with our data. These include the Hypopharyngeal Patient
Care Evaluation (a survey by American College of Sur-
geons)3based on over 2,000 patients across the United
States, a population-based study from Holland5and a 228
patient series from Germany.6The remainder of the liter-
ature consists of cohort studies from single institutions
contaminated by selection bias that tend to have either
small numbers of cases or patients from prior decades that
may not be relevant today.
MATERIALS AND METHODS
A provincial cancer database was created by the Division of
Cancer Care and Epidemiology (DCCE) (formerly the Radiation
Oncology Research Unit [RORU]) in 1995 consisting of linked
longitudinal data from multiple sources on all patients diagnosed
with cancer across Ontario from 1982. The analysis of this data
has resulted in numerous publications, abstracts, posters, and
presentations.7The RORU/DCCE database is updated biannu-
ally. The sources include:
1. The Ontario Cancer Registry (OCR): The OCR is a population-
based tumor registry operated by Cancer Care Ontario. It
consists of linked data on all patients with cancer including
demographic information from the cancer centers, all pathol-
ogy reports of cancer from all hospitals and laboratories, all
hospital separations from the Canadian Institute of Health
Information (CIHI) when a diagnosis of cancer included and
death information from the Ontario Registrar General.
2. Cancer center databases: Each of the Cancer Care Ontario
treatment centers have a common electronic database, devel-
oped in 1985, that collects further demographic information as
well as treatment summaries for all RT and chemotherapy.
The Princess Margaret Hospital in Toronto has a similar but
independent system.
3. Statistics Canada: Statistics Canada collects socioeconomic
status (SES) data as part of Census Canada. These data in-
clude information such as median household income (MHI) by
census units.
4. CIHI: The CIHI source files on all hospital separations pro-
vides dates of admission and discharge, up to five procedure
codes and 16 discharge diagnoses.
From the OCR, we identified 891 cases of squamous cell
carcinoma of the hypopharynx across Ontario diagnosed from
January 1, 1990 to December 31, 1999. Of the 891 cases there
were 35 duplicate cases leaving 856 patients. Specific clinical
information was requested and received from all Cancer Care
Ontario treatment centers and the Princess Margaret Hospital
for the variables to describe patients, tumors, treatments, and
clinical course including cause of death for the 856 patients.
Eleven identified patients that were not seen at a regional cancer
center were also included based on requested hospital records.
The charts of 50% (random) of the patients from the largest center
(126 were not reviewed) and the charts of all patients from other
eight cancer treatment centers were abstracted (total 595).
Twelve percent of the original patients coded as cancer of the
hypopharynxintheOCRwererejectedforerrorsofsite[supraglottic
larynx (47), oropharynx (35), unknown primary (14), other (7)].
Eight patients were rejected due to insufficient information (extent
of disease, comorbid illness, treatment, lack of follow-up) and 24
patients who did not meet criteria of the study (no biopsy report,
nonsquamous cell carcinoma histology, and treatment out of prov-
ince) were rejected. The principal investigator (S.H.) reviewed all
clinical staging, all surgical treatments and cause of death. Missing
data such as CT scan reports and operative notes was requested and
obtained on 57% of cases through hospitals and physicians.
We used MHI as an estimate of SES.8The MHI is the
median income of the general population in the patient’s neigh-
borhood (census enumeration unit) and is based on Census Can-
ada 1991 or 1995, depending on date of date of diagnosis, linked
to postal codes. MHI was compared using five quintiles with
equal numbers of patients based on all cancer patients across
Ontario and was compared the populations by quintile for each
major cancer site.
Laryngoscope 118: August 2008Hall et al.: Squamous Cell Carcinoma of the Hypopharynx
1363

Page 3

We could not use Tumor, Node, Metastases (TNM) catego-
ries as stated on the charts as they were frequently missing,
outdated, or inconsistent. To establish anatomic stage, the refer-
ring letters, consultation letters, operative biopsy reports, and
imaging were used. The elements of anatomic stage were based
on clinical examination and imaging only. For this study, with the
known poor reliability of clinical examination of neck masses,
imaging was used when available for N category. N2B category
was used for multiple nodes regardless of size and N3 category for
individual nodes over 6 cm. A record by an otolaryngologist/head
and neck surgeon or an experienced head and neck radiation
oncologist stating any vocal cord dysfunction or imaging report-
ing vocal cord fixation was used to identify vocal cord paralysis. In
keeping with the Surveillance Epidemiology and End Results
Program Code Manual9for ambiguous terms on extent of disease,
we identified cartilage invasion by the specific terms (invasion to,
onto, appears to, and compatible with invasion) and any cartilage
invasion was recorded as positive. The T and N categories were
then generated based on the TNM manual10noting specific rules
such as cartilage invasion in pyriform cancers. TNM site assign-
ment was based on the location of at least 60% of tumor and
category assignment followed TNM general rules such as using
the less advanced stage when there was doubt about the correct
assignment. Resectability was established by the principle inves-
tigator (S.H.) based on either a statement on the chart by the
consultant head and neck surgeon or the information including
clinical examinations and imaging. Resectability refers to initial
surgical treatment reflecting either the extent of the primary or
neck disease and did not include functional status, comorbid
illness, or patient preference.
To establish comorbidity we used the Cumulative Illness
Rating Scale (CIRS)11and the Adult Comormidity Evaluation
(ACE)12based on information on all pretreatment assessments.
The CIRS is a multi-item summative predictive index consisting
of 13 domains (cardiovascular, respiratory, vascular, hematology,
upper gastrointestinal, lower gastrointestinal, eye, ear, nose, and
throat, liver, renal, genital-urinary, musculoskeletal, central ner-
vous system, psychiatric), each with a 0 to 4 severity scale. The
individual system severity scales are single item category scales
that are summed to give a total numerical index to represent
medical burden as a continuous variable. For the scale, zero mean
no impairment, two means an impairment that interferes with
everyday life and requires treatment, and four mean life-
threatening illness with poor prognosis. We used the CIRS-G
upgraded version, as per Miller et al.13,14We also used the ACE.
The ACE-27 was reported by Piccirillo et al.12in 2004 and is
based on the previous Kaplan-Feinstein classification. The
ACE-27 has 27 domains within 12 systems (cardiovascular, re-
spiratory, gastrointestinal, renal, endocrine, central nervous sys-
tem, psychiatric, rheumatology, immunological, cancer, sub-
stance abuse, obesity) and grades diseases/conditions from 0 to 3
(none to severe) according to the severity and prognostic impact.
The criteria for scoring are explicit in the manual. The highest
ranked ailment determines the final score (0–3) unless there are
two scores of 2 when a 3 is designated, thus creating an ordinal
index. These two indexes are complementary as they count dif-
ferent illnesses, have different severity criteria and use different
scoring methods so tend to identify different comorbid illnesses,
different severity of comorbid illnesses and different patients.15
For example, in the CIRS all cancers are listed with the disease
site, whereas in the ACE-27 cancers are a separate domain.
Neither the CIRS nor ACE-27 has a specific domain for alcoholism,
which is known to be an independent prognostic factor in head and
neck oncology.16Based on the overall structure of the Kaplan-
Feinstein classification,17we created a scale consisting of no prob-
lem, a drinking problem (daily alcohol in excess) past or
present, complications (medical or mental) of drinking, or
severe (known delirium tremens or severe portal hyperten-
sion). We also created a scale for oral hygiene based on consul-
tations and dental assessments.
Performance status was assessed using the Eastern Coop-
erative Oncology Group (ECOG) scale augmented by abstracted
information based on the items and domains of the functional
status portion of the index of co-existent illness.18
We classified treatment into four groups: surgery ? postop-
erative RT, RT ? surgery for residual disease, palliative RT, and
no active treatment. Because of small numbers of cases (n ? 22),
chemotherapy with RT was combined with RT regardless of pro-
tocol as long as the intent was cure. Curative RT was defined
based on dose not the stated intent and was defined as minimum
50 Gy. No active treatment included patients with delay of first
treatment over 4 months postdiagnosis and all non-RT palliative
treatments. A first recurrence had to be preceded by either a
“disease free statement on the chart by the attending oncologist”
or clear margins at the time of surgery. Patients with residual
disease after RT did not have a disease free statement.
Using the chart information, requested further information
and recorded cause of death based on the Ontario Death Registry,
we assigned causes of death as either dead of hypopharynx cancer
or dead of another cause. Patients who died with active progress-
ing disease were assigned dead of cancer regardless of actual
cause (aspiration pneumonia, pulmonary embolism, etc). For this
study, patients who died within a month of treatment were as-
signed the cause of death secondary to their cancer. Patients who
were cancer-free but subsequently died with no clinical informa-
tion on cause were assigned “dead of another cause—cause not
determined” (n ? 51), and lost patients with no information on
vital status were assigned “status missing” (n ? 2).
RESULTS
The Patients
The distribution of age, sex, marital status, employ-
ment status, and MHI are found on Table I. The average age
was 66.4, 61% were married and 80% were either unem-
ployed, retired or on disability. Figure 1 compares the quin-
tiles of MHI for other common cancer sites across Ontario
and demonstrates that patients with hypopharyngeal cancer
are overall among the poorest cancer patients in Ontario.
The most common presenting symptoms were dys-
phagia (53%), hoarseness (39%), neck mass (37%), weight
loss (36%), sore throat (34%), otalgia (30%), and hemopty-
sis (5.5%).
Table II presents the data for indicators of the overall
health of the patient population including the ACE comor-
bidity scores. Figure 2 presents the comorbidity scores of
the CIRS (continuous variable). The distributions of the
comorbidity scores reflect a group of patients with signif-
icant co-existing illnesses with 50% of patients having
associated diseases with at least moderate decompensa-
tion. ECOG performance status is an assessment of the
overall current health of the patient and includes both
symptoms and the impact of relevant comorbid illnesses.
Although one third of patients were well, almost 40% had
a significant reduction in performance status. We comple-
mented the ECOG results and the comorbidity data by
abstracting the functional status section of the index of
co-existent illness to assess the impact of the comorbid
illnesses on the patient’s life. The most common systems
that compromised daily living were respiratory, cardiac,
peripheral vascular disease, and neurological disease with
Laryngoscope 118: August 2008
1364
Hall et al.: Squamous Cell Carcinoma of the Hypopharynx

Page 4

the latter having the highest number of most severe scores
(stroke, dementia, depression). Similar to other studies we
found the rate of alcohol abuse to be very high with over
70% of the patients having a history of heavy alcohol use.
The recording of dental/oral health is a unique feature of this
study. The presence and quality of the information varied
from center to center and some records included dental con-
sultations. Thirty percent of patients had full dentures and
of those with teeth 25% had poor to bad oral health. Seventy-
two patients had previous cancers—especially notable are
the 16 patients with previous laryngeal cancer treated with
RT. Three patients had more than one previous cancer.
Table III presents an overview of types and degrees of
the comorbidities in the patient group. As noted in the
Methods section, the two scales create different illness
groups and identify different patients with different ill-
ness severities. Using the criteria for the CIRS, the most
common diseased systems were respiratory (includes
smoking and chronic obstructive pulmonary disease),
hepatobiliary (alcohol abuse), and gastrointestinal (peptic
ulcer disease). The respiratory system also had the high-
est incidence of patients with more severe disease (num-
ber of patients with scores ?1). Using the criteria of the
ACE-27, the most commonly involved system was cardiac
(includes heart, peripheral vascular disease, and hyper-
tension) followed by respiratory and other cancers. The
incidence of diabetes was 8.4%. The respiratory system
had the greatest number of patients with moderate de-
compensation (score ? 2) and the group “other cancers”
represented the greatest number of severe decompensa-
tion (score ? 3). Co-existing disease was present in three
or more systems in over 60% of cases.
Across the eight geographic regions of the province
the descriptive variables that were statistically signifi-
cantly different were the MHI (P ? .0001), the CIRS index
of comorbidity (P ? .0001), ECOG (P ? .048), marital
status (P ? .029), and oral hygiene (P ? .0001). The
differences in MHI reflect the higher incomes in areas of
TABLE I.
Case Mix: Patient Variables—1.
Variablen Percent
Gender
Male
Female
Age (yrs)
?50
51–60
61–70
71–80
?80
Marital status
Married
Not
Employed
Yes
No
MHI (quintiles)
Q1 (lowest income)
Q2
Q3
Q4
Q5
483
112
81.2
18.8
28
156
229
125
57
4.7
26.2
38.4
21.1
9.6
228
358
39
61
120
466
20
79.6
173
133
105
86
95
29.2
22.3
17.6
14.2
15.9
MHI ? median household income.
Median household income by cancer site in Ontario
% of Pts in Income Quintile by Cancer Site
10.0
12.5
15.0
17.5
20.0
22.5
25.0
27.5
12345
Hypopharynx
Prostate
Breast
GYN
GI
H&N
Lung
% of ptsCancer Site
Esophagus
Fig. 1. Demonstrates the percent of patients for
each quintile of median household income for
each cancer site and shows that, overall, pa-
tients with cancer of the hypopharynx are the
poorest of all cancer patients.
Laryngoscope 118: August 2008 Hall et al.: Squamous Cell Carcinoma of the Hypopharynx
1365

Page 5

Toronto and Ottawa versus the lower income of the two
most rural regions of the province. One of the two regions
with the lowest incomes also had the highest comorbidity
scores and second highest ECOG scores; however, the
overall the correlation between ECOG and comorbidity
was moderate (Pearson ? 0.43).
The Extent of Disease
The distributions for subsite, histological differentia-
tion, TNM categories, TNM stage groups, and resectabil-
ity are found on Table IV. Overall, there were 4% stage 1,
13% stage 2, 27% stage 3, and 57.4% stage 4 with 19.5% of
cases deemed unresectable for surgery as the initial treat-
ment. There were no statistically significant differences
across the eight geographic regions of Ontario in the dis-
tribution of T category, N category, TNM stage, or
resectability.
The most common combination of T and N categories
was T3N2 (13.4%). The combinations of T2N0, T2N2, and
T3N0 accounted for greater than 10% each and with T3N2
accounted for 50% of cases. T4N0, T2N1, and T4N2 made
up 8% to 10% of cases each and the remainder of the
combinations represented less than 5% each.
The Clinical Course
Residual disease and recurrence. Overall, 112
(22.6%) of the 494 treated for cure had residual disease at the
end of initial treatment and of the 382 who had a disease-
free statement on the chart after initial treatment, 49.8%
Fig. 2. Presents the distribution of scores for the Cumulative Illness
Rating Scale and demonstrates the high incidence of comorbidity in
the patient population.
TABLE II.
Case Mix: Patient Variables—2.
Variablen Percent
ACE-27
0
1
2
3
ECOG
0
1
2
?3
Alcohol
NA
A drinking problem
Complications
Dental health
Dentures yes
Good
Fair
Bad
Poor
Previous cancers
Larynx
Lung
GI
Other H&N
Other
152
145
171
122
25.8
24.5
29.1
20.9
219
150
191
35
36.9
25.2
32.1
5.9
171
400
24
28.9
67.1
4.0
164
187
20
36
70
59.9*
6.4*
11.5*
22.4*
16
8
17
3
27
22.5†
11.3†
23.9†
4.2†
38†
ACE ? Adult Comormidity Evaluation; ECOG ? Eastern Cooperative
Oncology Group; NA ? nondrinker or not started; GI ? gastrointestinal;
H&N ? head and neck.
*Percent of those with dentures.
†Percent of the 71 with a previous cancer.
TABLE III.
Comparison by Disease System of the Frequency of the Raw
Scores for the CIRS and ACE-27 Indexes of Comorbidity.
Highest ScoreCIRSACE-27
System
Cardiac
Respiratory
Gastrointestinal
Neurological
Renal
Musculoskeletal
Miscellaneous
Hepato-biliary
Other cancers
Diabetes
Peripheral
Genitourinary
Alcohol
1
37
53
70
25
6
38
42
168
x
x
18
29
x
2
14
257
52
21
3
27
223
28
34
412
86
148
17
25
3
1
9
138
37
52
42
7
3
1
8
1
0
5
x
17
2
x
x
5
13
6
3
4
11
2
x
x
21
4
x
313
9
3
232
48
232
x
x
95
26
x
12
24
44
x
x
25
1
x
21
x
33
41
x
x
21
39
x
31
7
x
x
33
CIRS ? Cumulative Illness Rating Scale; ACE ? Adult Comormidity
Evaluation.
Laryngoscope 118: August 2008
1366
Hall et al.: Squamous Cell Carcinoma of the Hypopharynx