Long-term Quality of Life for Surgical and Nonsurgical Treatment of Head and Neck Cancer

Objective  To compare the long-term, health-related quality-of-life outcomes in patients with advanced head and neck cancer (HNC) treated with surgery and postoperative radiation therapy (SRT) or concurrent chemotherapy and radiation therapy (CRT).

Design  Matched-pair study comparing patients with advanced HNC treated with SRT or CRT at least 12 months after treatment. Patients completed 2 validated surveys addressing HNC-specific outcomes and depressive symptoms and provided information on employment and tobacco and alcohol use. Results for the 2 groups were compared using paired-sample t test and χ² analysis.

Setting  University-based study.

Patients  Patients with stage III or IV squamous cell carcinoma of the oropharynx, hypopharynx, and larynx who underwent SRT or received CRT.

Main Outcome Measures  Head and neck cancer–specific health-related quality of life from the Head and Neck Cancer Inventory and level of depressive symptoms from the Beck Depression Inventory.

Results  The matching process resulted in 27 patients in each treatment group. The HNC-specific domain scores (with higher scores representing better outcomes) for CRT vs SRT were eating, 37.8 vs 40.8 (P = .69); speech, 65.1 vs 56.0 (P = .23); aesthetics, 80.3 vs 69.2 (P = .14); and social disruption, 69.7 vs 70.6 (P = .90). Overall health-related quality of life was 64.0 with SRT and 55.0 with CRT (P = .142). For the Beck Depression Inventory (with higher scores representing worse outcomes), patients who underwent SRT had a mean score of 9.6 compared with 11.6 for patients who received CRT (P = .42).

Conclusion  As nonsurgical means of treating HNC have become more aggressive and surgical techniques have become more focused on function preservation and rehabilitation, the overall health-related quality of life resulting from these different approaches is similar.

The use of systemic chemotherapy to treat head and neck cancer (HNC) has increased substantially over the last 15 years. The sensitivity of squamous cell carcinoma (SCC) of the upper aerodigestive tract to several chemotherapeutic agents has been clearly demonstrated.¹ Chemotherapy has been used in conjunction with radiation therapy in the management of HNC in a variety of schemes, including induction, concurrent, intra-arterial, adjuvant (postdefinitive treatment), and, most recently, concurrent with postoperative radiation therapy.^2,3 Intra-arterial chemotherapy has demonstrated efficacy. However, this type of regimen is labor intensive and not routinely available.⁴ Adjuvant chemotherapy has not demonstrated a survival advantage, but it appears to decrease the incidence of distant metastasis.⁵ Induction (neoadjuvant) chemotherapy followed by radiation therapy has not been shown to confer a survival advantage.^5,6 However, it was through the use of induction chemotherapy in organ preservation studies that the use of chemotherapy and radiation therapy as a curative alternative to surgery and postoperative radiation therapy (SRT) became widely recognized.⁷ Four large meta-analyses and many individual randomized trials have demonstrated a survival advantage of 8% to 12% in patients with HNC when concurrent chemotherapy is given with radiation therapy (CRT) compared with radiation therapy alone.^8-11 These results have defined the most promising role for chemotherapy in the treatment of HNC.

Within the upper aerodigestive tract, the nasopharynx, oropharynx, hypopharynx, and larynx have been the sites most extensively studied. In the oropharynx, CRT may allow avoidance of surgery, particularly for tongue base tumors, which may have a detrimental effect on swallowing. For advanced SCC of the hypopharynx and larynx that would require a total laryngectomy, organ preservation may be achieved with CRT.^1,7

These aggressive CRT regimens, which may include more intensified altered fractionation irradiation, are associated with a significant increase in acute and late toxic adverse effects compared with conventionally administered radiation therapy alone. In the acute setting, these toxic adverse effects are similar to those anticipated for chemotherapy and include nausea and a hematopoietic toxic reaction in addition to often severe mucositis and other less frequently seen toxic effects.^12-14 The long-term problems center on compromised oral functioning, swallowing, soft tissue fibrosis, and related pain or discomfort.^13,15 Owing to the scarcity of data and heterogeneity of the protocols that have been reported, a clear picture of the long-term toxic effects associated with the use of CRT is unavailable. This deficiency in the literature was pointed out by the authors of all of the major meta-analyses evaluating chemotherapy and irradiation in the treatment of HNC.^8-11,16

Equally unfortunate is the lack of long-term functional status and quality-of-life (QOL) data from studies using CRT to treat HNC. There is clear evidence that patients receiving CRT have substantial problems with swallowing, in both the short and long terms.¹⁷ Pain, depression, aspiration, radionecrosis of the mandible, fibrosis, neuralgias, and dental problems may also result from CRT. However, the incidence and potential effect of these problems on functional status and QOL have not been rigorously evaluated.¹¹

Over the last 15 to 20 years, as CRT was emerging as a viable option for managing advanced HNC, substantial changes have also occurred in the surgical management of this disease. In the contemporary management of HNC the disfiguring and shoulder debilitating radical neck dissection is rarely performed.¹⁸ Conservation surgical procedures for the larynx with voice preservation and speech rehabilitation techniques are commonly available. Substantial progress in the reconstruction of the tongue, mandible, pharynx, and larynx has also been made.^19-22 Most of the major advances in the surgical treatment of HNC during the past 20 years have been directed primarily at improving functional status and QOL without compromising oncologic effectiveness.

To our knowledge, no randomized trials comparing CRT with SRT have been conducted. However, the common assumption remains that functional status and QOL are better with nonsurgical than with surgical approaches to manage advanced HNC. A growing body of evidence has demonstrated that although organ sacrifice may have a detrimental functional effect in a given area, patients are often rehabilitated or adapt much better than was anticipated and overall long-term QOL is frequently maintained.^23-28 In addition, it is now more commonly recognized that preservation of a nonfunctional organ is of little value.⁷ It is our hypothesis that as nonsurgical regimens become more aggressive and toxic and surgical interventions become less disfiguring and more function sparing, the functional status and QOL resulting from these 2 alternative treatment approaches will be similar in their effect on overall health-related QOL (HRQOL), with smaller differences seen in specific areas of functioning.

Patients in this study were drawn from those with an advanced-stage (III or IV) primary SCC of the oropharynx, hypopharynx, or larynx who were examined at the University of Iowa Hospital and Clinics, Iowa City, between January 1, 1991, and June 30, 2002. Patients in the surgical group underwent SRT and patients in the nonsurgical group received CRT. The treatment for each patient was based on a tumor board discussion of the best potential oncologic and HRQOL outcomes possible with the methods available at the time the patient received treatment.

This study used a matched-pair design. Patients who received CRT were paired with patients who underwent SRT on the basis of site (oropharynx or hypopharynx/larynx), age at presentation (within 10 years), and diagnostic year (within 2 years). Stage was not a criterion for matching, because all participants had advanced disease.

The method of accruing patients involved a 2-step process designed to provide the longest possible follow-up. In the first step, the University of Iowa Hospital and Clinics cancer registry was used to identify eligible patients who were alive at last contact at a minimum of 12 months after diagnosis. These patients were mailed the HRQOL surveys and a standardized form requesting information about employment and tobacco and alcohol use. In the second step, additional eligible patients who did not respond to the mailing were extracted from the Department of Otolaryngology’s ongoing longitudinal Outcomes Assessment Study (OAS) if they had provided 12-month, posttreatment HRQOL data while participating in that project. Patients who responded to the survey were matched with each other, and patients who were extracted from the OAS were matched with each other. In accord with the guidelines of the University of Iowa College of Medicine Human Subjects Investigational Review Board, participants in the mailing study were advised that returning the surveys was their acknowledgment of the informed consent provided in the cover letter, and participants in the OAS signed an informed consent document at the time of their enrollment.

The HRQOL outcomes of interest were speech, eating, aesthetics, social disruption, and depressive symptoms. Two validated surveys, the Head and Neck Cancer Inventory (HNCI)²⁹ and the Beck Depression Inventory (BDI),³⁰ were used to collect these data. The BDI is a 21-item survey that captures information about symptoms of depression using ordinal scale responses ranging from 0 to 3. Studies have provided normative results categorized as normal (0-9), mild (10-20), moderate (21-30), and severe (≥31) depressive symptoms.³¹ The HNCI is a 30-item survey that measures HNC-specific outcomes in the areas of speech, eating, aesthetics, and social disruption. The items cover functional ratings (how well the patient is functioning) and attitudinal ratings (how satisfied the patient is with a given functional level). The ordinal scale responses range from 1 to 5. The instrument also contains 1 item that addresses overall QOL. Items in which 1 represents the best outcome are inverted; thus, 5 represents the best outcome for all items. The 8 speech and eating items that have valid “not applicable” responses (because of a laryngectomy or because of receiving nothing by mouth) are coded as 1, the worst outcome. Domain scores are transformed to a scale of 0 to 100 for ease of interpretation.

Paired-sample t tests and χ² analyses were performed to determine whether the differences between the 2 treatment groups were statistically significant. To determine whether the differences between the means of these 2 treatment groups were clinically relevant, they were compared with previously derived clinically important differences (CIDs) for the HNCI and BDI scores. Table 1 summarizes the differences in BDI and HNCI domain scores that represent small, intermediate, and large CIDs.³²

In addition to demonstrating the mean scores for the 2 treatment groups, the percentage of patients who received CRT and who underwent SRT with favorable and unfavorable HNC-specific outcomes are reported. This categorization is accomplished by aggregating the HNCI scores into high (favorable), intermediate, and low (unfavorable) groups. High, intermediate, and low score groupings were assigned on the basis of a previous analysis of average scores for patients within various anchor health states relevant to the individual domains.³²

The first step of the accrual process that extracted eligible patients from the University of Iowa Hospital and Clinics cancer registry identified 148 patients (43 who received CRT, 105 who underwent SRT) who then received the surveys in the mail. Sixty-eight (20 who received CRT, 48 who underwent SRT) returned them for a response rate of 46%. The second step for accruing patients identified 70 eligible OAS participants (16 who received CRT, 54 who underwent SRT) who had provided 12-month HRQOL data but were excluded from the mailing. The first matching process involving patients who responded to the mailing resulted in 11 pairs. The second matching process involving patients who had taken part in the OAS longitudinal study resulted in an additional 16 pairs, bringing the total number of pairs to 27. Case-mix data for the 27 pairs of patients in the 2 treatment groups are given in Table 2. No significant differences in the case-mix data for the 2 groups were found.

In the comparison of mean HNC-specific scores between the 2 treatment groups (Table 3), higher HNCI scores represent better outcomes. The mean scores were similar for the SRT and CRT treatment groups in social disruption (70.6 and 69.7, respectively; P = .90) and eating (40.8 and 37.8, respectively; P = .69). The CRT group had higher scores in the speech domain (65.1 vs 56.0 for SRT, P = .23) and the aesthetics domain (80.3 vs 69.2 for SRT, P = .14). Although the differences in these HNC-specific domain scores were not statistically significant, the differences between the speech and the aesthetics domain scores for these 2 treatment groups were above the cutoff score for a small CID (Table 1). The mean overall QOL score (which represents a single item on the HNCI) for these 2 groups was 55.0 for CRT and 64.0 for SRT (P = .14). This difference was above the benchmark value for a small CID.

Unlike the HNCI scores, higher mean BDI scores represent worse outcomes (ie, higher levels of depressive symptoms). The CRT group mean BDI score of 11.6 was higher than the SRT group mean of 9.6. Although these scores were not statistically significantly different (P = .42), the difference was above the cutoff for a small CID for depressive symptoms (Table 1).

The distributions of HRQOL scores in Table 4 provide information about the proportion of patients within these treatment groups with outcomes in the low, intermediate, and high ranges. The distribution of low, intermediate, and high social disruption scores was fairly equivalent for the 2 treatment groups, but a substantially higher percentage of patients in the CRT group had eating scores in the low range (50.0% vs 34.6% for SRT). A substantially higher percentage of patients in the CRT group had scores in the high range for speech (55.6% vs 37.1% for SRT) and aesthetics (80.8% vs 63.0% for SRT). In the area of depressive symptoms, the percentages of patients with normal scores were fairly equivalent across the 2 treatment groups. No patients who underwent SRT reported severe levels of depressive symptoms compared with the 13.6% of patients who received CRT.

Figures 1, 2, and 3 illustrate average HNCI domain scores for patients with HNC in various anchor health states relevant to 3 of the domains. These scores are compared with the HNCI domain scores for patients in the CRT and SRT groups.³² The mean speech scores for both groups of patients fell between the laryngeal and alaryngeal anchor health states (Figure 1). The eating domain scores of the 2 treatment groups were less than the mean scores for patients who had an edentulous diet (ie, a diet without imposed restrictions that an edentulous patient can consume) (Figure 2). The social disruption scores for both groups (related to the clinical anchor of reported pain) were aligned with the average scores of patients who were experiencing mild pain (Figure 3).

A study of this size (27 pairs with 0.80 power, r = 0.70) would be unable to detect a statistically significant difference that was fewer than approximately 12 points for all domains except social disruption, which was approximately 8 points. None of the mean differences between the 2 treatment groups was this large.

The combination of radiation therapy with concurrently administered chemotherapy, primarily platinum based, has been seen to have a curative potential for many patients with advanced SCC of the oropharynx, hypopharynx, and larynx.¹¹ These treatment regimens are particularly attractive for patients in whom the alternative treatment would involve surgical resection of a large portion of the tongue base, oropharynx, hypopharynx, or larynx. The oncologic efficacy of CRT in this population in comparison with radiation therapy alone has been well documented.^8-11 However, little detailed information is available on the long-term HRQOL outcomes after CRT to treat advanced HNC or on any randomized trials comparing CRT with SRT. Despite this, the presumption persists that posttreatment QOL is generally better with CRT than with SRT.³³

Many patient, disease, and treatment factors will influence the ultimate long-term QOL in these patients. Health status outcomes are often referred to globally as QOL outcomes. However, they should more precisely be separated into objective, that is, functional or performance status outcomes (what an individual can do) and subjective QOL outcomes (evaluations of the way an individual perceives and reacts to what he or she is capable of doing).²⁶ Health status outcomes are multidimensional and include several areas or domains. It is possible that a single health care intervention will have positive effects in one area and negative effects in another.^32,34 The circumstances that determine how a person perceives or is affected by performance status (QOL), within a specific area of functioning, vary substantially between individuals, are subject to adaptation over time, and exert a variable influence on overall QOL for different individuals.^23,25,28 In addition, the general assumption that nonoperative intervention will uniformly lead to superior QOL also fails to account for the potential for a nonfunctioning but anatomically preserved organ; the possibility exists that there may be health states with a preserved organ (eg, chondronecrosis, chronic aspiration) that are less desirable than not having the organ⁷; and there are the possibilities of having advances in ablative surgical techniques, surgical reconstruction, and rehabilitation after surgery that preserve and restore function.

In this study we evaluated health status outcomes in several domains as well as global QOL. Differences in health status outcomes were evaluated based on CIDs as well as statistical significance. None of the measured outcomes showed a statistically significant difference between the 2 groups. However, several of the outcomes did demonstrate differences that were of the magnitude of a small CID.^32,35

Patients in the SRT group demonstrated worse speech outcomes than did patients in the CRT group; these outcomes were of a magnitude equal to a small CID, consistent with the likelihood that many had laryngeal ablative surgery. However, this difference was not carried over to the overall QOL, a measure in which the SRT group had a superior score. This is similar to other research findings that have demonstrated preservation of overall QOL despite the loss of laryngeal function.^23,27,28,34 Preservation of laryngeal speech after CRT is a real expectation for many patients who do not require salvage surgery.³⁶ Our previous work has shown that there is a large clinically significant difference in speech outcomes between patients who have laryngeal speech and those who do not.³² The preservation of overall QOL in those patients who have had laryngeal ablative surgery has been attributed to the available voice (communication) rehabilitation techniques and adaptation to methods of communication other than laryngeal speech.^23,28,37

In the domain of eating, the results were essentially identical for both groups. On the basis of several case series that evaluated eating and swallowing after CRT, it seems as if this treatment exerts a much more negative effect in this domain than in the speech domain.^17,34,36 Surgery on the tongue base and pharynx with laryngeal preservation may create problems for the preservation of a safe swallow. However, patients who have had a total laryngectomy no longer have the potential for aspiration. In these patients, as long as a conduit of satisfactory caliber is maintained, swallowing is not substantially affected. The long-term effects of CRT on the elasticity of the pharynx and upper esophagus create problems not only related to stricture but also to aspiration.^12,13,15 The potential loss of dentition, xerostomia, persistent mucosal tenderness, and soft tissue fibrosis after radiation therapy would in large measure be common to both groups. As seen in Figure 1, patients in the SRT and CRT groups are functioning at the level of patients with an edentulous diet. It is likely that a number of patients in both groups require at least supplemental tube feedings, and as Terrell et al³⁸ have demonstrated, the need for a feeding tube has a substantial negative effect on QOL related to eating and overall QOL.

The domain of social disruption captures information related to discomfort, pain, and altered social function. The 2 treatment groups demonstrated similar results in this domain. As with the eating domain, many of the deficits encountered will be common to both groups, such as eating in social situations, fatigue, discomfort related to either of the treatments, and problems related to partial or total edentulation.

The SRT group performed better in the areas of depressive symptoms and on the item evaluating overall QOL. A possible explanation for this relates to the definition of QOL as a measure of the difference between what patients want and expect and what they have.²⁶ It is possible that, in general, the CRT group anticipated less treatment effect on their lives than the SRT group did. It may be conceptually easier to prepare patients for the mechanical deficits after surgery than the less concrete difficulties that can be related to CRT. Thus, the lower overall QOL scores and greater depressive symptoms may be related to the unmet expectations that avoiding surgery would leave them with few posttreatment sequelae. Use of a feeding tube has a substantial negative effect on QOL in patients with HNC,³⁸ and a number of patients who received CRT required a long-term supplemental tube feeding.

Although differences in speech and aesthetics favored CRT and differences in depressive symptoms and overall QOL favored SRT, the differences were of the magnitude of a small CID and were not statistically significant. Overall, results in the 2 groups were remarkably similar. This is graphically evident in Figures 1, 2 and 3, which show the average scores for both treatment groups in the speech, eating, and social disruption domains and their corresponding health states.

A high percentage of patients in the CRT and SRT groups in this study reported relatively good HRQOL outcomes. It is possible that these patients may not have had as favorable a result with the alternative treatment. Several patients in each group also reported poor HRQOL outcomes. The key questions to be addressed in future work center on defining pretreatment indicators that will identify patients who are better candidates for a particular treatment method on the basis of anticipated survival and HRQOL outcomes.

This study was composed of a relatively high performing cohort of patients with HNC, both oncologically and in a more global social context. The patients in this study were willing to complete a questionnaire and had enjoyed at least 1 year of survival despite advanced disease. Although the matching process ensured that similar patients were compared, this article does not address HRQOL issues in poorer performing patients. This potential bias or lack of information should be recognized by the reader. The power calculation demonstrated that a mean difference of approximately 12 in the domain scores would have been required to show a statistically significant difference between groups. With more patients this number would have been lower, increasing the chance of demonstrating a statistically significant difference. However, conclusions drawn from the clinically important differences between scores would likely not change. Finally, the outcome data compared between the groups in this study are cross-sectional, representing 1 point in time, and do not consider the functioning of the patients throughout the period immediately after treatment or the baseline functioning of the patients.

This study supports the previous findings that, despite deficits in various areas of functioning, overall QOL is often preserved for many patients with HNC who have undergone SRT or received CRT. Although the assumption that preserving an organ will uniformly result in a higher QOL seems reasonable, the complexities of human adjustment and the multitude of potential treatment effects, from both surgery-based or nonoperative intervention, render this assumption invalid for many patients. Selection of treatment should be tailored to the individual, the tumor, and the most likely results, but it should not be based on the assumption that surgery will uniformly be a greater insult to function and QOL than nonsurgical intervention. This work supports the hypothesis that as nonsurgical interventions become more aggressive and surgery-based interventions become less disruptive to function, the health status outcomes for patients receiving CRT or undergoing SRT may favor one or another area or domain, but overall the outcomes will have more similarities than differences. The similarities will result from the common effects related to radiation therapy, similar overall oral and swallowing problems, and similar levels of social disruption related to a variety of factors.

Correspondence: Gerry F. Funk, MD, Department of Otolaryngology–Head and Neck Surgery, Pomerantz Family Pavilion, Room 21200, University of Iowa Hospitals and Clinics, 200 Hawkins Dr, Iowa City, IA 52242-1093 (gerry-funk@uiowa.edu).

Submitted for Publication: February 15, 2005; final revision received March 28, 2005; accepted May 2, 2005.

Financial Disclosure: None.

Funding/Support: This study was supported in part by grant RO1CA106908-01 from the National Institutes of Health through the Office of Cancer Survivorship, Bethesda, Md.

Previous Presentation: This study was presented at the annual meeting of the American Head and Neck Society; May 5, 2003; Nashville, Tenn.