[Skip to Content]
[Skip to Content Landing]
Figure 1.
One-Year Overall Survival as a Function of Time to Recurrence and Symptom Severity Stage
One-Year Overall Survival as a Function of Time to Recurrence and Symptom Severity Stage

Numerators are the number of survivors at 1 year, denominators are the number of patients at risk, and values within parentheses are the 1-year overall survival rates.

Figure 2.
One-Year Overall Survival as a Function of Functional Severity Staging System Stage and rTNM Stage
One-Year Overall Survival as a Function of Functional Severity Staging System Stage and rTNM Stage

Numerators are the number of survivors at 1 year, denominators are the number of patients at risk, and values within parentheses are the 1-year overall survival rates.

Figure 3.
Kaplan-Meier Analysis of Overall Survival for rTNM Stage, Functional Severity Staging System Stage, and Clinical Severity Staging System Stage
Kaplan-Meier Analysis of Overall Survival for rTNM Stage, Functional Severity Staging System Stage, and Clinical Severity Staging System Stage

A, P < .001; log-rank test, 22.1. B, P < .001; log-rank test, 44.4. C, P < .001; log-rank test, 53.7.

Table 1.  
Univariate Analysis of Demographics and Clinical Variablesa
Univariate Analysis of Demographics and Clinical Variablesa
Table 2.  
Univariate Analysis of Staging Systems
Univariate Analysis of Staging Systems
1.
Tanvetyanon  T, Padhya  T, McCaffrey  J,  et al.  Prognostic factors for survival after salvage reirradiation of head and neck cancer.  J Clin Oncol. 2009;27(12):1983-1991. doi:10.1200/JCO.2008.20.0691PubMedGoogle ScholarCrossref
2.
Parkin  DM, Bray  F, Ferlay  J, Pisani  P.  Global cancer statistics, 2002.  CA Cancer J Clin. 2005;55(2):74-108. doi:10.3322/canjclin.55.2.74PubMedGoogle ScholarCrossref
3.
Edge  SB, Byrd  DR, Compton  CC, Fritz  AG, Greene  FL, Trotti  A, eds.  AJCC Cancer Staging Manual. 7th ed. New York, NY: Springer; 2010.
4.
Piccirillo  JF.  Purposes, problems, and proposals for progress in cancer staging.  Arch Otolaryngol Head Neck Surg. 1995;121(2):145-149. doi:10.1001/archotol.1995.01890020009003PubMedGoogle ScholarCrossref
5.
Göllnitz  I, Inhestern  J, Wendt  TG,  et al.  Role of comorbidity on outcome of head and neck cancer: a population-based study in Thuringia, Germany.  Cancer Med. 2016;5(11):3260-3271. doi:10.1002/cam4.882PubMedGoogle ScholarCrossref
6.
Feinstein  AR, Pritchett  JA, Schimpff  CR.  The epidemiology of cancer therapy, IV: the extraction of data from medical records.  Arch Intern Med. 1969;123(5):571-590. doi:10.1001/archinte.1969.00300150089013PubMedGoogle ScholarCrossref
7.
Piccirillo  JF, Wells  CK, Sasaki  CT, Feinstein  AR.  New clinical severity staging system for cancer of the larynx: five-year survival rates.  Ann Otol Rhinol Laryngol. 1994;103(2):83-92. doi:10.1177/000348949410300201PubMedGoogle ScholarCrossref
8.
Pugliano  FA, Piccirillo  JF, Zequeira  MR,  et al.  Clinical-severity staging system for oropharyngeal cancer: five-year survival rates.  Arch Otolaryngol Head Neck Surg. 1997;123(10):1118-1124. doi:10.1001/archotol.1997.01900100094013PubMedGoogle ScholarCrossref
9.
Pugliano  FA, Piccirillo  JF, Zequeira  MR, Fredrickson  JM, Perez  CA, Simpson  JR.  Clinical-severity staging system for oral cavity cancer: five-year survival rates.  Otolaryngol Head Neck Surg. 1999;120(1):38-45. doi:10.1016/S0194-5998(99)70367-0PubMedGoogle ScholarCrossref
10.
Ribeiro  KC, Kowalski  LP, Latorre  MR.  Impact of comorbidity, symptoms, and patients’ characteristics on the prognosis of oral carcinomas.  Arch Otolaryngol Head Neck Surg. 2000;126(9):1079-1085. doi:10.1001/archotol.126.9.1079PubMedGoogle ScholarCrossref
11.
Reyes-Gibby  CC, Anderson  KO, Merriman  KW, Todd  KH, Shete  SS, Hanna  EY.  Survival patterns in squamous cell carcinoma of the head and neck: pain as an independent prognostic factor for survival.  J Pain. 2014;15(10):1015-1022. doi:10.1016/j.jpain.2014.07.003PubMedGoogle ScholarCrossref
12.
Feinstein  AR.  Symptoms as an index of biological behaviour and prognosis in human cancer.  Nature. 1966;209(5020):241-245. doi:10.1038/209241a0PubMedGoogle ScholarCrossref
13.
Feinstein  AR, Pritchett  JA, Schimpff  CR.  The epidemiology of cancer therapy, 3: the management of imperfect data.  Arch Intern Med. 1969;123(4):448-461. doi:10.1001/archinte.1969.00300140094023PubMedGoogle ScholarCrossref
14.
Feinstein  AR, Pritchett  JA, Schimpff  CR.  The epidemiology of cancer therapy, II: the clinical course: data, decisions, and temporal demarcations.  Arch Intern Med. 1969;123(3):323-344. doi:10.1001/archinte.1969.00300130105014PubMedGoogle ScholarCrossref
15.
Feinstein  AR.  The pre-therapeutic classification of co-morbidity in chronic disease.  J Chronic Dis. 1970;23(7):455-468. doi:10.1016/0021-9681(70)90054-8PubMedGoogle ScholarCrossref
16.
Piccirillo  JF, Tierney  RM, Costas  I, Grove  L, Spitznagel  EL  Jr.  Prognostic importance of comorbidity in a hospital-based cancer registry.  JAMA. 2004;291(20):2441-2447. doi:10.1001/jama.291.20.2441PubMedGoogle ScholarCrossref
17.
Feinstein  AR.  Multivariable Analysis: An Introduction. New Haven, CT: Yale University Press; 1996. doi:10.2307/j.ctt2250wks
18.
Feinstein  AR, Wells  CK.  A clinical-severity staging system for patients with lung cancer.  Medicine (Baltimore). 1990;69(1):1-33. doi:10.1097/00005792-199001000-00001PubMedGoogle ScholarCrossref
19.
Walter  SD, Feinstein  AR, Wells  CK.  A comparison of multivariable mathematical methods for predicting survival, II: statistical selection of prognostic variables.  J Clin Epidemiol. 1990;43(4):349-359. doi:10.1016/0895-4356(90)90121-5PubMedGoogle ScholarCrossref
20.
Feinstein  AR.  Clinical biostatistics, XVI: the process of prognostic stratification, 2.  Clin Pharmacol Ther. 1972;13(4):609-624. doi:10.1002/cpt1972134609PubMedGoogle ScholarCrossref
21.
Hosmer  DW, Lemeshow  S. Applied Logistic Regression. 2nd ed. New York, NY: Wiley; 2000. Wiley Series in Probability and Statistics.
22.
Nguyen  TV, Yueh  B.  Weight loss predicts mortality after recurrent oral cavity and oropharyngeal carcinomas.  Cancer. 2002;95(3):553-562. doi:10.1002/cncr.10711PubMedGoogle ScholarCrossref
23.
Yueh  B, Feinstein  AR, Weaver  EM, Sasaki  CT, Concato  J.  Prognostic staging system for recurrent, persistent, and second primary cancers of the oral cavity and oropharynx.  Arch Otolaryngol Head Neck Surg. 1998;124(9):975-981. doi:10.1001/archotol.124.9.975PubMedGoogle ScholarCrossref
24.
Lacy  PD, Spitznagel  EL  Jr, Piccirillo  JF.  Development of a new staging system for recurrent oral cavity and oropharyngeal squamous cell carcinoma.  Cancer. 1999;86(8):1387-1395. doi:10.1002/(SICI)1097-0142(19991015)86:8<1387::AID-CNCR2>3.0.CO;2-MPubMedGoogle ScholarCrossref
25.
Asano  T, Yamada  S, Fujii  T,  et al.  The Charlson age comorbidity index predicts prognosis in patients with resected pancreatic cancer.  Int J Surg. 2017;39:169-175. doi:10.1016/j.ijsu.2017.01.115PubMedGoogle ScholarCrossref
26.
Guldberg  TL, Christensen  S, Zachariae  R, Jensen  AB.  Prognostic factors in early breast cancer associated with body mass index, physical functioning, physical activity, and comorbidity: data from a nationwide Danish cohort.  Breast Cancer Res Treat. 2017;162(1):159-167. doi:10.1007/s10549-016-4099-yPubMedGoogle ScholarCrossref
27.
Voutsadakis  IA.  Obesity and diabetes as prognostic factors in patients with colorectal cancer.  Diabetes Metab Syndr. 2017;11(Suppl 1):S109-S114.PubMedGoogle Scholar
28.
Berrino  F, Gatta  G; EUROCARE Working Group.  Variation in survival of patients with head and neck cancer in Europe by the site of origin of the tumours.  Eur J Cancer. 1998;34(14 Spec No):2154-2161. doi:10.1016/S0959-8049(98)00328-1PubMedGoogle ScholarCrossref
29.
Boysen  ME, Zätterström  UK, Evensen  JF.  Self-reported symptoms to monitor recurrent head and neck cancer: analysis of 1,678 cases.  Anticancer Res. 2016;36(6):2849-2854.PubMedGoogle Scholar
30.
Matoscevic  K, Graf  N, Pezier  TF, Huber  GF.  Success of salvage treatment: a critical appraisal of salvage rates for different subsites of HNSCC.  Otolaryngol Head Neck Surg. 2014;151(3):454-461. doi:10.1177/0194599814535183PubMedGoogle ScholarCrossref
31.
Daly  ME, Le  QT, Maxim  PG,  et al.  Intensity-modulated radiotherapy in the treatment of oropharyngeal cancer: clinical outcomes and patterns of failure.  Int J Radiat Oncol Biol Phys. 2010;76(5):1339-1346. doi:10.1016/j.ijrobp.2009.04.006PubMedGoogle ScholarCrossref
32.
Huang  SH, Perez-Ordonez  B, Weinreb  I,  et al.  Natural course of distant metastases following radiotherapy or chemoradiotherapy in HPV-related oropharyngeal cancer.  Oral Oncol. 2013;49(1):79-85. doi:10.1016/j.oraloncology.2012.07.015PubMedGoogle ScholarCrossref
Original Investigation
August 2018

Association of Symptoms and Clinical Findings With Anticipated Outcomes in Patients With Recurrent Head and Neck Cancer

Author Affiliations
  • 1Department of Otolaryngology–Head & Neck Surgery, Washington University School of Medicine in St Louis, St Louis, Missouri
  • 2Editor, JAMA Otolaryngology–Head & Neck Surgery
JAMA Otolaryngol Head Neck Surg. 2018;144(8):738-745. doi:10.1001/jamaoto.2018.1230
Key Points

Question  Are symptoms and clinical findings at the time of diagnosis of recurrent head and neck cancer associated with disease outcomes?

Findings  In this cohort study of 196 patients, 1-year overall survival was 58.2% (114 of 196 patients). The clinical variables of time to recurrence and symptom severity stage were combined with rTNM stage to create a new 3-category staging system; the discriminative power of this new composite staging was better than that of the American Joint Committee on Cancer classification.

Meaning  Symptoms and clinical variables may be associated with anticipation of outcomes in patients with recurrent head and neck cancer.

Abstract

Importance  Despite advances in treatment over the last decades, recurrent head and neck cancer continues to have a poor prognosis. Prognostic accuracy may help in patient counseling.

Objective  To explore whether symptoms and clinical variables can predict prognosis in the setting of recurrent head and neck cancer.

Design, Setting, and Participants  In this retrospective cohort study, patients treated for head and neck cancer with curative intent at Siteman Cancer Center in St Louis, Missouri (a tertiary cancer center) between January 1, 2007, and December 31, 2014, were reviewed. The dates of data analysis were October 2016 to June 2017. Patients who developed a recurrent cancer were included, with 196 patients meeting inclusion criteria.

Main Outcomes and Measures  Symptoms and clinical findings at presentation of recurrence were recorded. Sequential sequestration and conjunctive consolidation (2 multivariable techniques) were used to create a composite staging system to predict 1-year overall survival (OS).

Results  Among 196 patients (mean [SD] age, 61 [11] years; 166 [84.7%] of white race/ethnicity; 76.5% male), 1-year OS was 58.2% (114 of 196 patients). Time to recurrence, symptom severity stage, and rTNM stage were consolidated into a 3-category Clinical Severity Staging System, with 1-year OS rates of 90.2% (95% CI, 82.7%-97.6%) for the 61 patients classified as A, 58.1% (95% CI, 47.7%-68.6%) for the 86 patients classified as B, and 18.4% (95% CI, 7.5%-29.2%) for the 49 patients classified as C. The discriminative power of the new composite staging was better than that of the American Joint Committee on Cancer classification (C = 0.79 vs C = 0.66).

Conclusions and Relevance  These findings suggest that clinical variables are associated with anticipated outcomes in patients with recurrent head and neck cancer.

Introduction

Despite advances in treatment over the last few decades, recurrent head and neck cancer continues to have a poor prognosis.1,2 When counseling patients with recurrent disease, it is important for clinicians to be able to give accurate and realistic estimates of prognosis. Prognostic estimates are drawn from previous patients with similar disease severity. To categorize patients into similar disease severity, staging systems for cancer were introduced. The most widely used staging system for head and neck cancer is the American Joint Committee on Cancer (AJCC) TNM staging classification.3 Staging systems help in treatment planning, aid in evaluation of treatment outcomes, and provide useful comparisons between the results of different patients and institutions. Most importantly, staging systems give an estimate of disease outcomes.4 The strength of the TNM staging system is its simplicity and relative accuracy of anticipated outcomes. However, the TNM staging system is a purely anatomical classification that describes the morphologic extension of primary tumor, lymph node involvement, and presence of distant metastatic disease and has multiple deficiencies.

First, the TNM staging system omits any description of the host. If the purpose of the staging system is to categorize patients into groups with similar outcomes and survival, exclusion of patient factors (eg, age, symptoms, and comorbidities) is counterintuitive. A young, otherwise healthy patient will have a different overall survival (OS) outcome than an elderly patient with advanced cardiovascular disease and the same morphologic cancer extent.5

Second, the TNM staging system fails to describe the biological behavior of the cancer. The classification implies that cancer has a linear expansion and predictable pattern of spread, from a small localized tumor, to regional metastasis, followed by distant metastasis. This is not always the case.

Third, specific challenges arise when staging recurrent cancers. Some of the defining structures may have already been removed at the time of initial treatment of the primary cancer (eg, laryngectomy for laryngeal cancer or the nodal basins after a neck dissection). The designation of disease extent thus becomes more challenging and inconsistent.

Symptoms, as a sign of disease aggressiveness, have been shown to increase the precision in the prediction of survival in various non–head and neck cancers.6 In previous studies7-11 focusing on primary head and neck cancer, the introduction of clinical variables improved survival prediction compared with TNM staging alone. We sought to identify whether inclusion of symptoms and other available clinical variables can improve the anticipation of survival outcomes compared with the current TNM staging system in patients with recurrent head and neck cancer.

Methods
Study Design and Population

This was a retrospective cohort study. Patients who developed a recurrent cancer after curatively intended treatment of squamous cell carcinoma in the oral cavity, oropharynx, larynx, or hypopharynx at Siteman Cancer Center in St Louis, Missouri (a tertiary cancer center), between January 1, 2007, and December 31, 2014, were identified. The dates of data analysis were October 2016 to June 2017. Recurrence was defined as a biopsy-proven tumor within the original tumor bed or regional or distant malignant cancer manifesting more than 3 months but less than 5 years after completion of curative treatment. Regional or distant disease was considered a recurrence if it either was biopsy proven with the same pathology as the primary tumor or was deemed to have originated from the primary tumor. Patients seen with a tumor within 3 months after definitive treatment were considered to have persistent disease and were excluded. Before commencement, this study was approved by the investigational review board at Washington University in St Louis. Because this was a retrospective medical record review, informed consent was waived.

Variables of the Study Population

Demographics included sex, race/ethnicity, age, overall comorbidity severity, risk exposure (eg, tobacco use and alcohol consumption history), time to recurrence, initial treatment modality, tumor site, histopathological grade, index tumor, anatomical type of recurrence, clinical and pathological TNM stage, and last known survival status. Additional variables extracted were TNM stage of recurrent cancer, p16 status of oropharyngeal cancers, symptoms related to the recurrence, amount of weight loss within 3 months before recurrence, and weight at the time of diagnosis.

The date of clinical or pathological diagnosis of recurrence was designated as zero time. To ensure reliable data collection, the first 10% of the medical records were reviewed by 2 of us (P.P. and J.L.).

Data Classification
Symptoms

Symptoms were classified into 3 distinct groups of locoregional, systemic, and distant.6,12-14 Local, systemic, and distant symptoms attributed to the recurrent cancer had to be consistent with recurrent disease, have no apparent association with treatment of the primary tumor or other coexistent disease, and persist up to the time of recurrence. Symptoms regarded as nonattributable were not recorded from our data extraction. Local and systemic symptom groups were further subdivided based on severity, resulting in a 6-category symptom severity score (asymptomatic, mild locoregional, severe locoregional, mild systemic, severe systemic, and distant).

Locoregional symptoms included local, perilocal, and regional symptoms, which can be attributed to the tumor at its primary site, to its association with adjacent structures, to the compromise of upper aerodigestive tract function, and to regional draining sites of the primary tumor. Examples include voice change, dysarthria, local pain, dysphagia, upper respiratory tract distress, local bleeding, facial swelling, cranial nerve palsy, xerostomia, dysgeusia, neck mass, and trismus. Patients seen with any of these symptoms at a prompted visit between their scheduled surveillance visits were considered to have severe locoregional disease, whereas patients seen at a regular follow-up visit were considered to have mild locoregional disease.

Systemic symptoms were manifestations at locations beyond the head and neck that were not directly related to metastasis. Where possible, weight loss was calculated as a percentage of body weight and classified as either none (<5%), mild (≥5% to <8%), or severe (≥8%). If a patient’s baseline weight was unknown, 4.5-kg (10-lb) loss was considered mild, and 6.8-kg (15-lb) loss was considered severe. If a patient’s weight loss was not provided but weight loss was the primary symptom, the weight loss was classified as mild. Other examples of systemic symptoms included fatigue, anorexia, dizziness, insomnia, nausea, and night sweats. Therefore, weight loss of 8% or higher or at least 6.8 kg (15 lb) was classified as severe systemic and all other as mild systemic.

Distant symptoms were defined as symptoms arising outside of the head and neck region and were representative of an objectively confirmed metastasis. Examples of distant symptoms included central nervous system disturbance or peripheral neurological signs, distant pain, extremity swelling, hemoptysis, and lower respiratory tract distress.

Comorbidity and Functional Status

Other diseases and conditions not related to the index cancer were defined as comorbidities.15 The Adult Comorbidity Evaluation 27 (ACE-27)16 comorbidity index is a validated instrument for patients with cancer. It includes a variety of individual comorbid ailments, each graded as 1 of 3 levels of comorbidity severity as follows: mild (grade 1), moderate (grade 2), and severe (grade 3). An overall comorbidity severity score was determined based on the grade of the highest-ranked single ailment; if 2 comorbid ailments of different body systems are graded as moderate (grade 2), the overall score is severe (grade 3).

Anatomical Staging of the Recurrent Tumor

The AJCC seventh edition TNM staging guidelines3 for head and neck cancer were followed for anatomical staging of the recurrent tumor. Data were extracted from radiological reports and from operative, tumor board, and other relevant physician notes.

p16 Status of Oropharyngeal Cancers

The p16 status of oropharyngeal cancers was recorded. To extract the p16 status, pathological reports and physician notes on the primary and recurrent cancer were examined.

Outcomes

Outcomes were recorded. The primary outcome measure was 1-year OS after zero time.

Data Analysis

Descriptive statistics were used to investigate the distribution of characteristics of interest in the sample. Univariate logistic regression was used to identify baseline demographic, clinical, and tumor features associated with 1-year OS. All significant variables were included in a multivariable logistic regression model. In addition, sequential sequestration and conjunctive consolidation17 were used to create a multivariable composite staging system prognostic of 1-year OS.

Sequential Sequestration

Because patients can have more than 1 symptom, we aimed to identify and separate any overlap between individual symptom classes. This resulted in the classification of symptoms into 6 distinct categories (asymptomatic, mild locoregional, severe locoregional, mild systemic, severe systemic, and distant). Sequential sequestration17 was then performed to confirm a prognostic gradient between patients based on the presence of the most severe symptom.

Conjunctive Consolidation

A challenge with the combination of multiple factors is the incremental increase in resulting bins or categories. Conjunctive consolidation17 is the cross-table analysis of the conjoined effect of 2 prognostic variables on the outcome of interest. The individual cells represent a group of patients with the same attributes for the 2 clinical variables.18,19 Adjacent cells are then consolidated according to statistical and clinical similarities. The consolidated cells create new categories within a new composite stage grouping. In this way, additional clinical variables can be combined while still keeping a small number of categories.

Kaplan-Meier Curves

Kaplan-Meier analysis was used to explore and compare OS between the categories of each staging system. The ability of each staging system to discriminate between patients who will be alive at 1 year after recurrence and those who will die within the first year of recurrence was assessed using C statistics. In logistic regression, the C statistic is the area under the receiver operating characteristic curve, and its value ranges from 0.5 to 1, with 0.5 indicating no better discrimination than could be achieved by chance alone and 1 indicating perfect discrimination.

Statistical Analysis

The software packages SAS version 9.4 (SAS Institute Inc) and IBM SPSS for Windows version 24 (SPSS Inc) were used for all statistical analyses. All statistical tests were 2-sided and evaluated at an α level of .05.

Results

The cohort of 196 patients included 150 men (76.5%) and 46 women (23.5%), with a mean (SD) age of 61 (11) years (Table 1). There were 166 patients (84.7%) of white race/ethnicity and 30 patients (15.3%) of nonwhite race/ethnicity. Fifty-two patients (26.5%) were seen with local recurrence, 65 (33.2%) with regional recurrence, and 79 (40.3%) with distant recurrence. One-year OS was 58.2% (114 of 196 patients).

Time to recurrence, treatment of the primary cancer, and initial and rTNM stage were associated with 1-year OS (Table 1). Sex, age, comorbidity, tobacco use, alcohol consumption, and primary site were not associated with 1-year OS.

Symptom Severity Stage

One-year OS as a function of the 6-category symptom severity classification demonstrated a strong linear prognostic gradient, with patients who had distant symptoms having worse survival (eTable in the Supplement). Because patients can have more than 1 symptom, sequential sequestration was then performed to classify patients based on the highest-severity symptom (eTable in the Supplement). An important result of the sequential sequestration process is that patients are only counted once, and patients are classified based on their highest-severity symptom; therefore, the number of patients assigned to the less-severe symptom severity categories is reduced compared with the results in the original cohort before any sequestration. To confirm the importance of classifying patients with multiple symptoms based on the highest-ranked severity symptom, patients with the highest-ranked symptom were sequentially removed from the cohort, and 1-year OS for the remaining patients is available in the eTable in the Supplement. With the removal of each successive symptom severity class, starting with the removal of patients with distant symptoms, 1-year OS increased from 58.2% to 70.4%.

Development of the Functional Severity Staging System

Time to recurrence and symptom severity stage were the 2 clinical variables that had the strongest association with 1-year OS, and they were combined through the process of conjunctive consolidation. In Figure 1, 1-year OS rates are shown for conjoined categories of time to recurrence and symptom severity stage. In general, within any category of time to recurrence, symptom severity stage defines unique prognostic gradients. In a similar way, within each category of symptom severity stage, time to recurrence defines unique prognostic gradients. When both variables show a gradient in the same direction, this is referred to as a double-gradient phenomenon, revealing the distinctive association of each variable with the other. Next, the individual cells with similar clinical characteristics and survival rates were consolidated to create the new 3-category Functional Severity Staging System (α, β, and γ). For example, patients with no or only mild locoregional recurrence more than 12 months after treatment and patients with severe locoregional (at a prompted visit) more than 24 months out are classified as α. One-year OS rates are 90.0% (95% CI, 81.7%-98.3%) for the 50 patients in α, 62.0% (95% CI, 52.0%-71.9%) for the 92 patients in β, and 22.2% (95% CI, 11.1%-33.3%) for the 54 patients in γ.

Development of the Clinical Severity Staging System

To incorporate the prognostic information provided by the morphologic extent of recurrence, a second conjunctive consolidation was performed to add rTNM to the Functional Severity Staging System (Figure 2). The categories of functional severity were α, β, and γ, as described. Within each category of the Functional Severity Staging System, rTNM defines unique prognostic gradients; similarly, within each category of rTNM, Functional Severity Staging System stage defines unique prognostic gradients (double gradient). The categories A, B, and C are the result of the consolidation of α, β, and γ with rTNM. One-year OS rates were 90.2% (95% CI, 82.7%-97.6%) for the 61 patients in A, 58.1% (95% CI, 47.7%-68.6% for the 86 patients in B, and 18.4% (95% CI, 7.5%-29.2%) for the 49 patients in C. Univariate analysis of each staging system is summarized in Table 2.

Quantitative Comparison of Staging Systems

We compared the performance of each staging system using the discriminative power (C statistic), survival gradient, and monotonicity of survival. These are well-established criteria for quantitative comparison of staging systems.20 Compared with rTNM staging, functional and clinical staging demonstrated greater discriminative power, as shown by their respective C statistics: rTNM stage (0.66; 95 % CI, 0.59-0.74), Functional Severity Staging System (0.77; 95% CI, 0.71-0.84), and Clinical Severity Staging System (0.79; 95% CI, 0.73-0.86). The range between 95% CIs for the 2 proposed staging systems exceeds the acceptable limits of a strong model (C > 0.7).21 The survival experience as defined by the categories of each of the staging systems is shown in Kaplan-Meier curves of OS in Figure 3. While the survival gradient at 1 year is similar between the functional and clinical staging systems, the monotonicity and separation of the survival curves (survival gradient) are much more evident in the clinical (71.8%) and functional (67.8%) stages than by rTNM stage (42.8%).

Discussion

This study found that the addition of symptom severity at the time of presentation of recurrence may significantly improve anticipation of disease outcomes beyond a system based on morphology alone. Within the study cohort, there is a subgroup with late recurrence and mild symptoms who had much better potential disease outcomes. We chose to combine the factors time to recurrence and symptom severity stage in a Functional Severity Staging System because they were independently significantly associated with 1-year OS. Both of these variables were also readily available at the time of recurrence. By combining those variables, we created a staging system that better stratified survival than the TNM classification. Furthermore, we consolidated this new staging system with the current rTNM system to create a Clinical Severity Staging System. This model marginally improved the discrimination. No other variable provided sufficient prognostic information to justify the added complexity of incorporation into the model. For a staging system to be meaningful and useful for clinicians, it also needs to be intuitive and easy to use.

Our data corroborate the findings by Nguyen and Yueh22 that weight loss is associated with outcomes in patients with recurrent head and neck cancer. Whereas weight loss and previous treatment were combined in their study, we instead incorporated the information on weight loss in our comprehensive symptom score. Various studies have used different amounts of weight loss to try to anticipate outcomes. In our study, we noted a difference in outcomes even with a modest amount of weight loss (ie, >2%), although this may have been an overlap outcome from severe local symptoms. In the study by Nguyen and Yueh,22 previous treatment was also associated with survival, which we noted in this study as well. In their study, recurrence after radiotherapy-based treatment was associated with worse outcomes.

In an earlier study by Yueh et al,23 a staging system for recurrent oral and oropharyngeal cancer was developed. In that study, muscular invasion was noted to have a significant association with survival and was incorporated into their model. Unfortunately, muscular invasion is not routinely noted in the pathology reports at Siteman Cancer Center. The main focus in the study by Yueh et al23 was also slightly different because they sought to explore differences between persistent and recurrent disease and new primary tumors. Their main finding that TNM performed poorly as a method to anticipate outcomes in patients with recurrent head and neck cancer was confirmed in our cohort. In another study, Lacy et al24 explored symptoms, clinical variables, and comorbidity as predictors of outcome in recurrent oral and oropharyngeal cancer. Initial TNM and extent of recurrence, as well as time to recurrence, were associated with 2-year OS. They combined initial TNM and extent of recurrence to create a new staging system. The OS in their cohort was 20%, reemphasizing the poor survival outcomes associated with recurrent head and neck cancer.

Comorbidity in various other cancers has been shown to influence OS.16,25-27 In a study by Ribeiro et al,10 the presence of comorbidity was associated with the OS in primary oral cancer and was combined with symptoms to create a new staging system. They used the Charlson Comorbidity Index; we instead used the ACE-27, a comorbidity index that was specifically designed for and validated in a population of patients with cancer. However, the presence of comorbidity was not associated with survival in our cohort. These findings are similar to the data by Yueh et al23 and by Nguyen and Yueh.22 We hypothesize that given the low OS in recurrent head and neck cancer the association of chronic disease with survival is low.

We combined oral, oropharyngeal, and laryngeal/hypopharyngeal cancer subsites in our cohort. Although these cancers behave slightly differently in a primary setting,28 the survival after recurrence appears to be reasonably similar.29,30 There was a slightly better OS in the group with oropharyngeal cancer; however, this was not clinically significant. Notably, our inception cohort came from a group of patients with a slightly lower recurrence rate than that reported in the literature.2,22,24 We believe that this difference might be driven by the lower recurrence rate in p16-positive oropharyngeal cancer. This may also have contributed to our higher rate of distant recurrent disease because these cancers have low local and regional recurrence rates.31,32 Subgroup analysis between the different subsites, as well as between the p16-positive and p16-negative groups within oropharyngeal cancer, showed no significant survival difference, justifying the further analysis as a combined group. Although inclusion of different subsites in our analysis may have contributed to heterogeneity, the results from the study may also be more generalizable to a comprehensive head and neck practice treating a wide range of patients with cancers of the upper aerodigestive tract.

Limitations

We acknowledge that there are several limitations to our study. This was a cohort study based on a retrospective medical record review, which is dependent on the scope and quality of the data in the medical records. Furthermore, the proposed staging needs to be externally validated in a different cohort to assess its true applicability in clinical practice. This could either be done in a prospective cohort or be performed with another retrospective cohort at another institution. Unfortunately, national databases do not report on symptoms at the time of recurrence.

Conclusions

Although recurrent head and neck cancer is associated with an overall poor OS, there is a subset of patients who experience more favorable outcomes. The Functional Severity Staging System, combining time to recurrence and symptom severity stage, may be a useful tool for clinicians when counseling patients with recurrent head and neck cancer.

Back to top
Article Information

Accepted for Publication: April 21, 2018.

Corresponding Author: Patrik Pipkorn, MD, Department of Otolaryngology–Head & Neck Surgery, Washington University School of Medicine in St Louis, 660 S Euclid Ave, Campus Box 8057, St Louis, MO 63110 (ppipkorn@wustl.edu).

Published Online: July 12, 2018. doi:10.1001/jamaoto.2018.1230

Author Contributions: Dr Pipkorn 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.

Concept and design: Pipkorn.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Pipkorn, Licata, Piccirillo.

Critical revision of the manuscript for important intellectual content: Pipkorn, Kallogjeri, Piccirillo.

Statistical analysis: Licata, Kallogjeri.

Supervision: Pipkorn, Piccirillo.

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

Disclaimer: Dr Piccirillo is the Editor of JAMA Otolaryngology–Head & Neck Surgery but was not involved in any of the decisions regarding review of the manuscript or its acceptance.

References
1.
Tanvetyanon  T, Padhya  T, McCaffrey  J,  et al.  Prognostic factors for survival after salvage reirradiation of head and neck cancer.  J Clin Oncol. 2009;27(12):1983-1991. doi:10.1200/JCO.2008.20.0691PubMedGoogle ScholarCrossref
2.
Parkin  DM, Bray  F, Ferlay  J, Pisani  P.  Global cancer statistics, 2002.  CA Cancer J Clin. 2005;55(2):74-108. doi:10.3322/canjclin.55.2.74PubMedGoogle ScholarCrossref
3.
Edge  SB, Byrd  DR, Compton  CC, Fritz  AG, Greene  FL, Trotti  A, eds.  AJCC Cancer Staging Manual. 7th ed. New York, NY: Springer; 2010.
4.
Piccirillo  JF.  Purposes, problems, and proposals for progress in cancer staging.  Arch Otolaryngol Head Neck Surg. 1995;121(2):145-149. doi:10.1001/archotol.1995.01890020009003PubMedGoogle ScholarCrossref
5.
Göllnitz  I, Inhestern  J, Wendt  TG,  et al.  Role of comorbidity on outcome of head and neck cancer: a population-based study in Thuringia, Germany.  Cancer Med. 2016;5(11):3260-3271. doi:10.1002/cam4.882PubMedGoogle ScholarCrossref
6.
Feinstein  AR, Pritchett  JA, Schimpff  CR.  The epidemiology of cancer therapy, IV: the extraction of data from medical records.  Arch Intern Med. 1969;123(5):571-590. doi:10.1001/archinte.1969.00300150089013PubMedGoogle ScholarCrossref
7.
Piccirillo  JF, Wells  CK, Sasaki  CT, Feinstein  AR.  New clinical severity staging system for cancer of the larynx: five-year survival rates.  Ann Otol Rhinol Laryngol. 1994;103(2):83-92. doi:10.1177/000348949410300201PubMedGoogle ScholarCrossref
8.
Pugliano  FA, Piccirillo  JF, Zequeira  MR,  et al.  Clinical-severity staging system for oropharyngeal cancer: five-year survival rates.  Arch Otolaryngol Head Neck Surg. 1997;123(10):1118-1124. doi:10.1001/archotol.1997.01900100094013PubMedGoogle ScholarCrossref
9.
Pugliano  FA, Piccirillo  JF, Zequeira  MR, Fredrickson  JM, Perez  CA, Simpson  JR.  Clinical-severity staging system for oral cavity cancer: five-year survival rates.  Otolaryngol Head Neck Surg. 1999;120(1):38-45. doi:10.1016/S0194-5998(99)70367-0PubMedGoogle ScholarCrossref
10.
Ribeiro  KC, Kowalski  LP, Latorre  MR.  Impact of comorbidity, symptoms, and patients’ characteristics on the prognosis of oral carcinomas.  Arch Otolaryngol Head Neck Surg. 2000;126(9):1079-1085. doi:10.1001/archotol.126.9.1079PubMedGoogle ScholarCrossref
11.
Reyes-Gibby  CC, Anderson  KO, Merriman  KW, Todd  KH, Shete  SS, Hanna  EY.  Survival patterns in squamous cell carcinoma of the head and neck: pain as an independent prognostic factor for survival.  J Pain. 2014;15(10):1015-1022. doi:10.1016/j.jpain.2014.07.003PubMedGoogle ScholarCrossref
12.
Feinstein  AR.  Symptoms as an index of biological behaviour and prognosis in human cancer.  Nature. 1966;209(5020):241-245. doi:10.1038/209241a0PubMedGoogle ScholarCrossref
13.
Feinstein  AR, Pritchett  JA, Schimpff  CR.  The epidemiology of cancer therapy, 3: the management of imperfect data.  Arch Intern Med. 1969;123(4):448-461. doi:10.1001/archinte.1969.00300140094023PubMedGoogle ScholarCrossref
14.
Feinstein  AR, Pritchett  JA, Schimpff  CR.  The epidemiology of cancer therapy, II: the clinical course: data, decisions, and temporal demarcations.  Arch Intern Med. 1969;123(3):323-344. doi:10.1001/archinte.1969.00300130105014PubMedGoogle ScholarCrossref
15.
Feinstein  AR.  The pre-therapeutic classification of co-morbidity in chronic disease.  J Chronic Dis. 1970;23(7):455-468. doi:10.1016/0021-9681(70)90054-8PubMedGoogle ScholarCrossref
16.
Piccirillo  JF, Tierney  RM, Costas  I, Grove  L, Spitznagel  EL  Jr.  Prognostic importance of comorbidity in a hospital-based cancer registry.  JAMA. 2004;291(20):2441-2447. doi:10.1001/jama.291.20.2441PubMedGoogle ScholarCrossref
17.
Feinstein  AR.  Multivariable Analysis: An Introduction. New Haven, CT: Yale University Press; 1996. doi:10.2307/j.ctt2250wks
18.
Feinstein  AR, Wells  CK.  A clinical-severity staging system for patients with lung cancer.  Medicine (Baltimore). 1990;69(1):1-33. doi:10.1097/00005792-199001000-00001PubMedGoogle ScholarCrossref
19.
Walter  SD, Feinstein  AR, Wells  CK.  A comparison of multivariable mathematical methods for predicting survival, II: statistical selection of prognostic variables.  J Clin Epidemiol. 1990;43(4):349-359. doi:10.1016/0895-4356(90)90121-5PubMedGoogle ScholarCrossref
20.
Feinstein  AR.  Clinical biostatistics, XVI: the process of prognostic stratification, 2.  Clin Pharmacol Ther. 1972;13(4):609-624. doi:10.1002/cpt1972134609PubMedGoogle ScholarCrossref
21.
Hosmer  DW, Lemeshow  S. Applied Logistic Regression. 2nd ed. New York, NY: Wiley; 2000. Wiley Series in Probability and Statistics.
22.
Nguyen  TV, Yueh  B.  Weight loss predicts mortality after recurrent oral cavity and oropharyngeal carcinomas.  Cancer. 2002;95(3):553-562. doi:10.1002/cncr.10711PubMedGoogle ScholarCrossref
23.
Yueh  B, Feinstein  AR, Weaver  EM, Sasaki  CT, Concato  J.  Prognostic staging system for recurrent, persistent, and second primary cancers of the oral cavity and oropharynx.  Arch Otolaryngol Head Neck Surg. 1998;124(9):975-981. doi:10.1001/archotol.124.9.975PubMedGoogle ScholarCrossref
24.
Lacy  PD, Spitznagel  EL  Jr, Piccirillo  JF.  Development of a new staging system for recurrent oral cavity and oropharyngeal squamous cell carcinoma.  Cancer. 1999;86(8):1387-1395. doi:10.1002/(SICI)1097-0142(19991015)86:8<1387::AID-CNCR2>3.0.CO;2-MPubMedGoogle ScholarCrossref
25.
Asano  T, Yamada  S, Fujii  T,  et al.  The Charlson age comorbidity index predicts prognosis in patients with resected pancreatic cancer.  Int J Surg. 2017;39:169-175. doi:10.1016/j.ijsu.2017.01.115PubMedGoogle ScholarCrossref
26.
Guldberg  TL, Christensen  S, Zachariae  R, Jensen  AB.  Prognostic factors in early breast cancer associated with body mass index, physical functioning, physical activity, and comorbidity: data from a nationwide Danish cohort.  Breast Cancer Res Treat. 2017;162(1):159-167. doi:10.1007/s10549-016-4099-yPubMedGoogle ScholarCrossref
27.
Voutsadakis  IA.  Obesity and diabetes as prognostic factors in patients with colorectal cancer.  Diabetes Metab Syndr. 2017;11(Suppl 1):S109-S114.PubMedGoogle Scholar
28.
Berrino  F, Gatta  G; EUROCARE Working Group.  Variation in survival of patients with head and neck cancer in Europe by the site of origin of the tumours.  Eur J Cancer. 1998;34(14 Spec No):2154-2161. doi:10.1016/S0959-8049(98)00328-1PubMedGoogle ScholarCrossref
29.
Boysen  ME, Zätterström  UK, Evensen  JF.  Self-reported symptoms to monitor recurrent head and neck cancer: analysis of 1,678 cases.  Anticancer Res. 2016;36(6):2849-2854.PubMedGoogle Scholar
30.
Matoscevic  K, Graf  N, Pezier  TF, Huber  GF.  Success of salvage treatment: a critical appraisal of salvage rates for different subsites of HNSCC.  Otolaryngol Head Neck Surg. 2014;151(3):454-461. doi:10.1177/0194599814535183PubMedGoogle ScholarCrossref
31.
Daly  ME, Le  QT, Maxim  PG,  et al.  Intensity-modulated radiotherapy in the treatment of oropharyngeal cancer: clinical outcomes and patterns of failure.  Int J Radiat Oncol Biol Phys. 2010;76(5):1339-1346. doi:10.1016/j.ijrobp.2009.04.006PubMedGoogle ScholarCrossref
32.
Huang  SH, Perez-Ordonez  B, Weinreb  I,  et al.  Natural course of distant metastases following radiotherapy or chemoradiotherapy in HPV-related oropharyngeal cancer.  Oral Oncol. 2013;49(1):79-85. doi:10.1016/j.oraloncology.2012.07.015PubMedGoogle ScholarCrossref
×