The article by Moore provides an example of much needed research evaluating clinical outcomes in head and neck oncology. Measuring the quality of life (QOL) of patients with head and neck cancer presents some unique challenges. First, head and neck cancer profoundly influences some of the most fundamental functions of life, including breathing, eating, and communication. Second, treatment of head and neck cancer does not always improve these functional deficits, and in many instances, the treatment itself results in further deterioration of these functions. Finally, "traditional" outcome measures (disease-free survival, overall survival, local and regional control, response rates) do not adequately assess the global impact of this disease and/or its treatment on patients' perception of life satisfaction.
Sometimes the treatments used for head and neck cancer, which include surgery, radiation, or chemotherapy (used as single modalities or in any combination), can result in disabilities that patients perceive as being worse than the untreated disease.[1,2] Therefore, there is a need for the use of validated instruments to quantify the impact of head and neck cancer and its treatment on patients' QOL.
Although interest in the incorporation of QOL outcome measures in head and neck cancer has recently increased, several questions remain unanswered: Which instrument should be used? Is a general, disease-specific, or site-specific instrument most appropriate in this patient population? Is head and neck cancer a single disease or a heterogeneous collection of the same disease at different subsites? Are certain instruments more specific to some treatment modalities than others? What is the most appropriate frequency and duration of QOL assessment in head and neck cancer patients? Who should administer the questionnaires? This discussion will explore some of the controversy surrounding these issues.
Which QOL Instrument to Use?
General vs Disease Specific--Health-related QOL instruments can be divided into general and disease-specific instruments. General measures assess the overall impact of patients' health status on their QOL. Examples of such instruments include the Sickness Impact Profile (SIP), the Quality of Well-Being Scale, and the Medical Outcome Study (MOS). General instruments employ detailed questionnaires, including numerous items that cover a broad spectrum of functional, physical, psychological, and social domains. Their main advantage is that they can be utilized across a broad range of patients, and allow for comparison of results across different diseases. The disadvantage of general instruments is their lack of responsiveness to the peculiar aspects of a particular disease process, such as cancer.
Disease-specific instruments are designed to evaluate in more depth the intricate features of a particular disease process. Examples of cancer-specific instruments include different performance status measures (Karnofsky, ECOG, AJCC), the Quality of Life Index, and the Functional Living Index-Cancer, or FLIC). These instruments are often more responsive to changes in cancer patients' health status over time than are general measures.
Site-Specific vs Modular Instruments--Although the diagnosis and treatment of cancer result in certain salient and well-recognized effects common to all cancer patients, the location of disease poses specific problems unique to that particular site. The head and neck region provides a very dramatic example of this concept. The devastating effects of head and neck cancer and/or its treatment on voice, speech, mastication, deglutition, sense of taste and/or smell, and facial appearance are peculiar to this patient population. Site-specific instruments are designed to be more responsive than cancer-specific instruments in measuring those difficulties that are most pronounced at a particular site. Examples of head and neck-specific instruments include the Performance Status Scale for Head and Neck Cancer Patients (PSS-HN) used in the Moore article, the University of Washington Head and Neck Questionnaire (UW QOL), and the Head and Neck Radiation Therapy Questionnaire (HNRQ).
Site-specific instruments may lack some of the broader assessment of the impact of cancer on patients' QOL. To combine the benefits of both approaches, some instruments have been designed with a measurement strategy called the "modular approach." In this approach, a core disease-specific instrument is used for all patients with the diagnosis of cancer, supplemented with a site-specific instrument that varies with the site of origin or method of treatment. Examples of such instruments are the Functional Assessment of Cancer Therapy (FACT)[reference 14 and personal communication, D. F. Cella, md, January 10, 1992] and the European Organization for Research and Treatment of Cancer (EORTC) Core Quality of Life Questionnaire, both of which have supplementary head and neck modules. This modular approach may prove to be ideal in measuring QOL in patients with head and neck cancer.
Is Head and Neck Cancer a Single Disease?
Patients with head and neck cancer have certain similarities. Most of these patients give a long-standing history of heavy tobacco and alcohol(Drug information on alcohol) consumption. These carcinogenic factors also predispose patients to common comorbidities, including chronic obstructive pulmonary disease, pancreatitis, cirrhosis, delirium tremens, and second malignancies, especially lung cancer. These comorbidities have a somewhat uniform impact on the QOL of head and neck cancer patients regardless of the primary site.
Conversely, the biologic behavior, and hence, outcome of head and neck cancer depend largely on the site of disease (eg, larynx, oral cavity, nasopharynx, and so on). Also, the disabilities that result from head and neck cancer or its treatment vary widely according to the site of involvement. The article by Moore presents an excellent example of measuring QOL in a specific subsite of a relatively homogeneous patient population with base of tongue cancer.
Treatment-Specific QOL Measures
Some QOL instruments are better suited to measuring the morbidity of surgical extirpation of head and neck cancer, while others are more sensitive to the morbidity associated with radiation and/or chemotherapy. For instance, the UW QOL includes domains that assess pain, disfigurement, chewing, swallowing, speech, shoulder disability, activity, recreation, and employment--all of which are profoundly affected by surgical ablation of head and neck cancer (eg, laryngectomy, glossectomy, mandibulectomy, neck dissection). On the other hand, the HNRQ contains items detailing the assessment of mucositis, skin reactions, xerostomia, nausea, vomiting, appetite, and energy level, among others. These items are more profoundly influenced by radiation and/or chemotherapy than by surgery.
In the article by Moore, the PSS-HN was chosen to assess the QOL of patients treated primarily with radiation therapy. The three domains detailed by the PSS-HN (eating in public, understandability of speech, and normalcy of diet) are more likely to be influenced by surgical ablation of head and neck cancer than by radiation therapy. The PSS-HN was initially validated in patients treated largely with surgery, and its results were highly responsive to the extent of surgical resection. This instrument does not specifically assess radiation-induced morbidity. This probably leads to underestimation of the effects of radiation on the QOL of these patients. It also makes comparison between the two modalities of treatment (surgery vs radiation) less meaningful. An instrument specifically designed to assess radiation-induced morbidity (eg, the HNRQ or EORTC) would have been more appropriate in a patient population treated primarily with radiation.
Frequency and Duration of QOL Assessment
Temporal factors greatly influence the results of QOL assessment. When measuring the acute morbidity of a disease process or its treatment, one should measure QOL frequently. In the case of radiation therapy or chemotherapy, QOL measurement probably should be done before treatment, at weekly or biweekly intervals during treatment, and for several weeks afterward.
Some therapeutic modalities have delayed effects (months or years) on QOL, and thus, require delayed measurement. If such a determination is not performed, the impact of this delayed morbidity may be missed. An example of this pitfall is the study by Harrison et al cited in the Moore article. Harrison et al used the PSS-HN in patients with base of tongue cancer treated with surgery and/or radiation. The radiation-therapy group had consistently better scores, and the authors recommended using radiation as a primary modality of therapy because the radiation-therapy group had survival rates equal to those of the surgery group but had better QOL.
Several caveats about these conclusions are worth mentioning. First, the previously mentioned limitation of the PSS-HN being a more surgery-sensitive than radia-
tion-sensitive instrument applies to the Harrison et al study as well as to the Moore study. Second, Harrison et al gave no indication of the extent of surgical resection. Finally, and perhaps most importantly, the Harrison et al study did not report long-term QOL scores.
These issues are crucial because before any modality, either surgical or radiotherapeutic, becomes accepted as the primary treatment for any particular type of head and neck cancer, its long-term effect on QOL should be known. For instance, Larson et al reported on long-term complications in 128 patients who had cancer of the oral cavity and oropharynx treated with radiation therapy alone and who were free of disease for at least 5 years. The study reported a 56.3% overall incidence of soft-tissue ulceration, osteoradionecrosis, or spontaneous fracture. Osteoradionecrosis occurred in 44 of 119 patients, and the incidence of this complication increased over time (42% within 2 years, 56% within 3 years, and 82% within 5 years). Out of these 44 patients, 18 required subsequent partial mandibulectomy or hemimandibulectomy.
This study illustrates the need for long-term QOL measurement especially when dealing with a treatment modality that has a relatively high rate of delayed complications.
Who Should Administer QOL Questionnaires?
Questionnaires on QOL can be self-reported (administered by the patients themselves) or interviewer-reported (administered by a physician, nurse, or other professional).
The PSS-HN used in the article by Moore is an example of an interviewer-administered questionnaire. The main advantage of these instruments is that they avoid vague or incomplete answers. Their main disadvantage is that the quality of data is highly interviewer-dependent. Possible areas of interviewer bias include their training and preconceptions about patients' demographics (age, sex, race, education, social class, and so forth), the disease process, or the treatment utilized. Several studies have demonstrated the poor correlation between patients' self-ratings and interviewer-generated ratings. Currently, patient self-reporting is generally the preferred method of obtaining QOL data.
The article by Moore describes the use of a well-defined QOL instrument in a homogeneous group of patients with base of tongue cancer treated with radiotherapy. Future studies in a similar patient population should perhaps consider an instrument that is more suited to the assessment of radiation-induced morbidity and its effect on QOL in such patients. Other factors that may enhance the ability to assess QOL in head and neck cancer include long-term reporting, patient-administered questionnaires, and instruments of the "modular" type that involve a core cancer-specific questionnaire with a head and neck specific module.