Estimation of an optimal external beam radiotherapy utilization rate for head and neck carcinoma

Authors

  • Geoff Delaney M.B.B.S., M.D.,

    Corresponding author
    1. Collaboration for Cancer Outcomes Research and Evaluation, Liverpool Hospital, Sydney, Australia
    • Collaboration for Cancer Outcomes Research and Evaluation, Liverpool Hospital, Locked Bag 7103, Liverpool, New South Wales, 1871, Australia
    Search for more papers by this author
    • Fax: (011) 61 2 9828 5299

  • Susannah Jacob M.B.B.S., M.D.,

    1. Collaboration for Cancer Outcomes Research and Evaluation, Liverpool Hospital, Sydney, Australia
    Search for more papers by this author
  • Michael Barton M.B.B.S.

    1. Collaboration for Cancer Outcomes Research and Evaluation, Liverpool Hospital, Sydney, Australia
    Search for more papers by this author

Abstract

BACKGROUND

Radiotherapy is used commonly in the treatment of patients with head and neck carcinoma. The benchmark radiotherapy utilization rates for head and neck carcinoma largely are unknown. The objective of the current study was to determine the optimal radiotherapy utilization rate for patients with head and neck carcinoma and to compare this optimal rate with actual utilization rates where actual utilization data were available.

METHODS

An optimal radiotherapy utilization tree was constructed that depicted all patients with head and neck carcinoma in whom radiotherapy was indicated according to evidence-based treatment guidelines. The proportions of patients with clinical attributes that indicated possible benefit from radiotherapy were obtained from epidemiological data and were inserted into the utilization tree. The optimal proportion of patients with carcinoma of the head and neck who should receive radiotherapy was calculated by merging the evidence-based recommendations with the epidemiological data in the tree. Optimal rates of radiotherapy utilization were compared with actual rates obtained from population-based studies.

RESULTS

Radiotherapy was indicated at some point during their illness in 74% of all patients with head and neck carcinoma. By subsite, the optimal radiotherapy utilization rates were oral cavity, 74%; lip, 20%; larynx, 100%; oropharynx, 100%; salivary gland, 87%; hypopharynx, 100%; nasopharynx, 100%; paranasal sinuses, 100%; and unknown squamous cell carcinoma of the head and neck, 90%. All treatment recommendations were based on Level III or IV evidence. Assessment of actual radiotherapy utilization rates indicated an increased use of radiotherapy over time for head and neck carcinoma. However, there also were some decreases in the use of radiotherapy for some carcinoma subsites over the past 20 years, despite the lower actual rates compared with the optimal rates. The reasons for these reductions in use were not identified.

CONCLUSIONS

The actual radiotherapy utilization rate for patients with head and neck carcinoma corresponded reasonably closely to the optimal rate for some populations but also identified some shortfalls for other patient groups. The results of this study provide a way of assessing shortfalls in radiotherapy. Cancer 2005. © 2005 American Cancer Society.

Radiotherapy is an essential treatment for malignant disease that contributes to the cure or palliation of many patients. Radiotherapy facilities have high initial capital costs, and their operation is staff-intensive. One of the benchmarks of the adequacy of radiotherapy service provision is the proportion of patients with new diagnoses of malignancies who receive at least one course of radiotherapy during the course of their illness (the radiotherapy utilization rate). Overall radiotherapy utilization rates vary substantially within Australia and internationally,1–6 with utilization rates ranging from 20% to 55% for all patients with newly diagnosed disease. It is important to derive an optimal rate of radiotherapy utilization that is based on the best available clinical and epidemiological evidence. Optimal radiotherapy utilization rates for breast carcinoma,7, 8 lung carcinoma,9, 10 gastrointestinal malignancies,11, 12 melanoma,13 gynecologic malignancies,14, 15 and prostate carcinoma16 have been reported previously. In this report, we present the radiotherapy utilization results for head and neck malignancies. The objectives of this study were to estimate the optimal proportion of all patients with head and neck carcinoma who, according to the best available evidence, should receive at least one course of external beam radiotherapy at some time during the course of their illness and to compare the optimal radiotherapy utilization rate for head and neck carcinoma with current practice.

MATERIALS AND METHODS

Indications for Radiotherapy

We defined an indication for external beam radiotherapy as a clinical situation in which radiotherapy is the treatment of choice on the basis of evidence that it will produce a better clinical outcome compared with the outcomes produced by other treatment modalities (including no treatment) and in which the patient is an appropriate candidate for radiotherapy in terms of performance status indicators and the presence or absence of comorbidities. The superiority of radiotherapy over other treatment options could be based on survival, local control, or toxicity profile. Because the objective of this investigation was to assess the optimal utilization rate for external beam radiotherapy, estimates for the use of brachytherapy were not included in the study.

Search Strategy and Selection Criteria

Currently, there are no national Australian treatment guidelines for the management of head and neck carcinoma. International guidelines for the management of head and neck carcinoma were identified. They are the United States National Cancer Institute Physician Data Query (PDQ) guidelines for head and neck carcinoma,17–24 the British Columbia Cancer Agency treatment guidelines,25–32 the National Comprehensive Cancer Network guidelines,33, 34 the guidelines issued by the British Association of Otorhinolaryngologists and Head and Neck Surgeons,35 and the European School of Oncology guidelines.36–38 The level of evidence that supported each recommendation for radiotherapy use was classified using the Australian National Health and Medical Research Council hierarchy of levels of evidence.39

Using the best evidence available, a list of indications for radiotherapy in the management of head and neck carcinoma was created, as shown in Table 1, and an optimal radiotherapy utilization tree was generated. In the radiotherapy utilization tree, each branch point represents an attribute (such as the stage of the tumor or whether clear surgical margins were present) that affects a radiotherapy treatment decision. Each terminal branch of the tree shows whether or not radiotherapy is recommended for individuals who have those particular clinical attributes.

Table 1. Head and Neck Cancer: Indications for Radiotherapy and Levels and Sources of Evidence
Outcome no. in treeClinical scenarioTreatment indicatedLevel of evidenceaReference(s)bProportion of all head and neck cancers
  • RT: radiotherapy; post-op: postoperative; ±: with or without; CT: chemotherapy; NCI/PDQ: National Cancer Institute/Physician Data Query; NCCN: National Comprehensive Cancer Network; BCCA: British Columbia Cancer Agency; FIGO: International Federation of Gynecology and Obstetrics; ABS: American Brachytherapy Society.

  • a

    Levels of evidence: Level I, systematic review of all relevant randomized studies; Level II, at least one properly conducted randomized trial; Level III: well designed controlled trials without randomization (includes trials with “pseudorandomization” or comparative studies); Level IV: case series. Adapted from the National Health and Medical Research Council's levels of evidence (National Health and Medical Research Council, 199839).

  • b

    Numbers in this column refer to the list of References.

1Oral cavity, Stages I–II, surgery, adverse pathologySurgery + post-op RTIV17, 25, 33, 35, 420.02
2Oral cavity, Stages I–II, surgery, no adverse pathology, locoregional recurrenceSurgery + post-op RTIV17, 250.01 (0.02)
4Oral cavity, Stages I–II, RTRadical RTIII17, 25, 33, 350.01
5Oral cavity, Stages III–IVSurgery + RTIII17, 25, 33, 350.15
6Lip, cosmetically excisable, locoregional recurrence RT +/− surgeryRT ± surgeryIV17, 330.02
8Lip, not cosmetically excisableRTIII17, 26, 33, 350.02
9Larynx, supraglottic, conservative surgery, locoregional recurrenceRTIV18, 33< 0.01
11Larynx, supraglottic, not suitable for larynx-preserving surgeryRTIII18, 30, 33, 35, 380.06
12Larynx, glottic, and subglottic, Stages I–II, RT appropriateRTIII18, 30, 33, 35, 380.07
14Larynx, glottic, and subglottic, Stage IIICT + RTIII18, 33, 380.03
15Larynx, glottic, and subglottic, Stage IVSurgery + post-op RT or RT ± CTIII18, 33, 35, 380.04
16OropharynxRT ± CT or Surgery + RTIII27, 33, 35, 360.08
17Salivary gland, Stages I–II, low grade, lymph node positivePost-op RTIV33, 35, 37< 0.01
18Salivary gland, Stages I–II, low grade, lymph node negative, locoregional recurrenceRadical or post-op RTIV22, 31, 33< 0.01
20Salivary gland, Stages I–II, high gradePost-op RTIV22, 31, 33, 350.03
21Salivary gland, Stages III–IVPost-op RTIV22, 31, 33, 350.02
22HypopharynxRadical or post-op RT ± CTIII20, 28, 33, 350.05
23Paranasal sinusRadical or post-op RTIII23, 33, 350.05
24NasopharynxCT + RT or RT aloneIII21, 29, 33, 350.04
25Unknown primary, N1–N2a, local or regional recurrenceRT alone or surgery + RTIV34, 35< 0.01
27Unknown primary, N2b–N3RT alone or surgery + RTIV24, 340.02
Proportion of all patients with head and neck cancer who have RT recommended   0.74 (74%)

An indication for radiotherapy may occur in the initial stages of treatment or may be delayed (for instance, in patients who develop a local or distant recurrence and who have not previously required radiotherapy as part of the original management). Patients who require external beam radiotherapy were counted only once, even if they had multiple indications for radiotherapy at different stages in their illness. This allowed comparison of the optimal rate with the actual radiotherapy utilization rate (defined as the number of patients treated by radiotherapy for the first time divided by the incidence of specific malignancies during a defined period).

Incidence Data

We collected epidemiological data on the proportion of patients who had attributes for which radiotherapy may be indicated and ranked the relative quality of epidemiological data using the hierarchy shown in Table 2. The source with the highest quality ranking was used to determine the incidence of each radiotherapy indication. Australian epidemiological data from Australian national and state cancer registries40, 41 were used whenever possible, because the results from this study will be used specifically to plan future radiotherapy facilities in Australia. When national data were unavailable, more specific data (such as those of state cancer registries) were used for information pertaining to tumor stage and pathology. Institutional data from published articles were used when higher levels of epidemiological data could not be obtained.42–57

Table 2. Head and Neck Cancer: The Incidence of Attributes Used to Define Indications for Radiotherapy
Population or subpopulation of interestAttributeProportion of the population with the attributeQuality of informationaReference(s)
  • AIHW: Australian Institute of Health and Welfare; SA: South Australia.

  • a

    Quality of information/hierarchy for epidemiological data: α: Australian national epidemiological data; β: Australian State Cancer Registry; γ: epidemiological data bases from other large international groups (e.g., Surveillance, Epidemiology, and End Results Program); δ: results from reports of a random sample from a population; ϵ: comprehensive multiinstitutional data base; ζ: comprehensive single-institutional data base; τ: multiinstitutional reports on selected groups (e.g., multiinstitutional clinical trials); λ: single-institutional reports on selected groups of patients (adapted from Sackett, 198959).

All registry cancersHead and neck cancers0.04αAIHW, 200140
All head and neck cancersOral cavity cancers0.28αAIHW, 200140
Oral cavity cancersStages I–II0.45βSA Cancer Registry, 200058
Oral cavity cancers, Stages I–IISurgery0.90αSA Cancer Registry, 200058
Oral cavity cancers, Stages I–II, surgeryAdverse pathology0.20ζJones et al., 199242
Oral cavity cancers, Stages I–II, surgery, no adverse pathologyLocoregional recurrence0.19ϵWolfensberger et al., 200157
All head and neck cancersLip cancer0.22αAIHW, 200140
Lip cancerCosmetically excisable0.89ζZitsch et al., 199544
Lip cancer, cosmetically excisableLocoregional recurrence0.09–0.14ζZitsch et al., 199544 and Rowe et al., 199245
All head and neck cancersLaryngeal cancer0.20αAIHW, 200140
Laryngeal cancerSupraglottic laryngeal cancer0.28ζSpitz et al., 198846
Supraglottic laryngeal cancerSuitable for larynx-preserving surgery0.16–0.26, 0.0ζLee et al., 199047 and Hinerman et al., 200248; see text
Supraglottic laryngeal cancer treated with larynx-preserving surgeryLocoregional recurrence0.16ζOrus et al., 200049
Laryngeal cancerStage I–II0.51βSA Cancer Registry, 200058
Laryngeal cancerStage III0.21βSA Cancer Registry, 200058
Laryngeal cancerStage IV0.28βSA Cancer Registry, 200058
All head and neck cancersCancer of the oropharynx0.08αAIHW, 200140
All head and neck cancersSalivary gland cancer0.06αAIHW, 200140
Salivary gland cancerStage I–II0.66–0.67ζCalerao et al, 199850; Spiro et al., 198951; and O'Brien et al., 198652
Salivary gland cancer, Stage I–IILow grade0.24–0.32ζSpiro et al, 198951; O'Brien et al., 198652; and North et al., 199053
Salivary gland cancer, low gradeLymph node positive0.05ζSpiro et al, 198951
Salivary gland cancer, low grade, negative marginsLocoregional recurrence0.26ζNorth et al., 199053
All head and neck cancersCancer of the hypopharynx0.05αAIHW, 200140
All head and neck cancersCancer of the paranasal sinus0.05αAIHW, 200140
All head and neck cancersNasopharyngeal cancer0.04αAIHW, 200140
All head and neck cancersUnknown primary (head and neck)0.02ζSinnathamby et al., 199754 and Gran et al., 200055
Unknown primary (head and neck)N1–N2a0.22–0.61ζSinnathamby et al., 199754 and Nguyen et al., 199456
Unknown primary (head and neck), N1–N2aLocal or regional recurrence0.54ζGran et al., 200055

An important source of data was the South Australian Network of Hospital-Based Cancer Registries.58 The registries in the network are based in major teaching hospitals and include data on patients attending the seven largest cancer centers in South Australia, which manage > 50% or the cancer cases in the state. The network data base includes information on disease stage that was not available in other state data bases.

We ranked the relative quality of epidemiological data from various sources, as shown in Table 2, using a modification of the hierarchy reported by Sackett.59 The hierarchy was modified to rank Australian data higher than data from other countries, because the results of this study will be used to plan future radiotherapy services in Australia. We used the highest ranking source to calculate the frequency of each radiotherapy indication.

Performance status data that were specific to patients with carcinoma of the head and neck could not be identified and, thus, were not incorporated into the tree. However, for most patients with locoregional disease, the performance status has little impact on the choice of treatment modality and may influence only the treatment intent. For instance, radiotherapy (with or without chemotherapy) may be appropriate in a patient who has an advanced head and neck primary tumor and a good performance status. The patient most probably still will undergo radiotherapy (but with palliative intent and in the absence of chemotherapy) if he has a poor performance status. Therefore, the omission of performance status from the tree should not make a large difference to the optimal radiotherapy utilization rate.

There will be some circumstances in which a patient with early-stage head and neck carcinoma who normally would undergo surgery, instead, will undergo radiation due to severe comorbidities, poor performance status, or patient choice. Similarly, there will be occasional patients with such poor performance status that no antitumor treatment is considered appropriate. These patients also are likely to be very rare. Omission of these patient groups from the radiotherapy utilization tree will not affect the overall optimal radiotherapy utilization rate significantly, because the numbers are comparatively small.

The issue of disease stage used in the tree was problematic, because the staging systems have changed over time, and published reports frequently do not state specifically which staging system was used. In addition, the incidence data were from differing periods; hence, different staging systems have been used. For the radiotherapy utilization trees, the treatment branches are based on the most recent staging system, i.e., the 1997 American Joint Committee on Cancer staging system.60 When old incidence data were used, the old staging system was applied despite the fact that the proportions in different stages may differ between the old and new staging systems. This was a pragmatic decision, because it is unlikely that these changes in the staging system would alter the proportion of a particular stage dramatically and, thus, it should have no significant effect on the optimal radiotherapy utilization rate.

Optimal Radiotherapy Utilization Rates

The optimal utilization rate was calculated by determining the incidence of each indication for radiotherapy. Summing these incidences yields the total optimal utilization rate for head and neck carcinoma. This optimal utilization rate was then compared according to tumor histology and disease stage with actual utilization data obtained from patterns-of-care studies.

The trees and the epidemiological data were sent for external review to a multidisciplinary panel of oncology experts (from surgical, medical oncology, radiation oncology, palliative care, and nursing backgrounds). The review process was overseen by an independent steering committee of general oncology experts that was convened by the National Cancer Control Initiative (Australia).

RESULTS

The optimal radiotherapy utilization tree for head and neck carcinoma is shown in Figure 1. Each branch of the tree signifies an attribute that affects a management decision (e.g., the stage of the tumor, whether surgical margins are clear, etc.). Above each branch in the radiotherapy utilization tree is a description of the specific attribute that has led to the treatment decision. Each number below the branch signifies the proportions of the attribute based on epidemiological data.

Figure 1.

Optimal radiotherapy (RT) utilization tree for patients with head and neck carcinoma.

There were 27 possible outcomes or end branches for this radiotherapy treatment tree. We identified 21 outcomes for which radiotherapy is indicated. All recommendations for radiotherapy from evidence-based guidelines were based on Level III or IV evidence, because there were no randomized data available comparing radiotherapy with other modalities. However, there are randomized data supporting the use of chemoradiation over radiation alone for patients with advanced laryngeal carcinoma.61 Table 1 identifies the clinical situations in which radiotherapy is recommended, the level of evidence, and the guideline or source of evidence for the recommendation. The outcome numbers in the table correspond to terminal nodes or outcome positions in the tree. The last column represents the incidence of each clinical indication for radiotherapy as a proportion of all patients with head and neck carcinoma. This is calculated by multiplying the incidences of all the branches leading up to the terminal branch. The sum of this column is the optimal radiotherapy utilization rate for head and neck carcinoma. This is calculated at 0.74, i.e., 74% of all patients with head and neck carcinoma should receive radiotherapy based on the best available evidence. Table 2 shows the epidemiological data corresponding to each branch point and the source of those data as well as the hierarchical level of the data obtained. Analysis of head and neck carcinoma shows that the optimal radiotherapy utilization rates for the various subsites were as follows: oral cavity, 74%; lip, 20%; larynx, 100%; oropharynx, 100%; salivary gland, 87%; hypopharynx, 100%; nasopharynx, 100%; paranasal sinuses, 100%; and metastatic squamous neck carcinoma of unknown origin, 90%.

Controversies in the Management of Head And Neck Carcinoma

In some instances, published treatment guidelines for head and neck carcinoma were not sufficiently specific in their recommendations to identify definite clinical or pathologic attributes as indications for radiotherapy. This may reflect the relative paucity of good-quality evidence that exists for head and neck carcinoma compared with other malignancies, such as breast carcinoma. Examples in which the guidelines were uncertain or vague and the strategies that were used to overcome this uncertainty are described below.

Treatment of patients with Stage I–II carcinoma of the oral cavity may consist of either surgery or radiotherapy, both of which are considered equally effective. Therefore, the actual treatment received may be affected by selection and referral biases. The British Columbian treatment guidelines25 recommend the use of radiotherapy alone for most patients with early-stage oral cavity carcinomas, citing excellent local control and functional results (Level III or IV evidence). The National Cancer Institute PDQ statement on lip and oral cavity carcinoma17 states that surgery or radiation therapy produce similar cure rates for patients with early-stage carcinoma (Stage I and II) of the oral cavity. The PDQ17 recommends that the choice of treatment should be dictated by the anticipated functional and cosmetic results of treatment and by the availability of specialist expertise and patient preference. In Australia, the majority of early-stage oral carcinomas are treated with surgery. Because there is no clear evidence to support one treatment modality over the other, the tree reflects current practice in Australia. The South Australia Hospital Cancer Registry reported that 90% of patients with Stage I–II carcinomas of the oral cavity underwent surgery,58 and this figure was used in the tree for the calculation of the optimal radiotherapy utilization rate. We have assumed that the other 10% of patients either were considered medically inoperable or refused surgery. Some of the surgical patients in the South Australia Hospital registry also received radiotherapy, presumably due to the presence of adverse pathologic features. For sensitivity analysis, the proportion of patients with Stage I–II disease who underwent surgery as primary therapy was modeled between 90% and 0% (i.e., at 1 extreme, 90% of patients underwent surgery; and, at the other extreme, all patients received radiation). The reason why 1 extreme was set at only 90% of patients undergoing surgery rather than 100% is that at least 10% of these patients are likely to be medically unfit to undergo major surgery. There is further discussion below on sensitivity analysis and the effect of this uncertainty on the final estimates.

A subset of patients with early-stage carcinoma of the oral cavity who undergo surgery may have adverse pathologic features that would warrant adjuvant radiotherapy. However, the specific pathologic features that would indicate adjuvant therapy are not specified in the evidence-based treatment guidelines. Jones et al.42 reported retrospectively on the recurrence patterns of 49 patients who underwent surgery for Stage I–II oral cavity carcinomas (tongue, retromolar trigone, floor of mouth, alveolar ridge, buccal mucosa, and hard palate). Pathologic factors that were associated with an unacceptably high locoregional recurrence rate were tumors invading > 5 mm in depth or tumors with positive margins. Patients who had these features represented ≈ 20% of the entire surgical group. The recurrence rate in patients who had no adverse pathology was not reported. Patients who develop a recurrence after surgery alone most likely will receive radiotherapy at the time of recurrence (either alone or combined with surgery).

The management of lip carcinoma also was controversial. The National Cancer Institute PDQ statement states that surgery and radiation therapy both are effective in the treatment of early-stage carcinomas (Stage I and II) of the lip and that the choice of treatment should be made based on the anticipated results (functional and cosmetic) of treatment and on the availability of experts. The size of the lesion would be the greatest determinant of the extent of resection and, hence, the greatest determinant of cosmetic and functional results. However, a size cut-off recommendation for surgery is not mentioned. For the purposes of the radiotherapy utilization tree, a tumor size of 3 cm was chosen arbitrarily as surgically excisable with a reasonable and acceptable cosmetic result based on expert opinion. Sensitivity analysis modeling was undertaken by varying the tumor cut-off size for surgery between 2 cm and 4 cm. All patients with lesions larger than the cut-off measurement were deemed to require either definitive radiotherapy or radiotherapy combined with surgery.

The management of Stage III laryngeal carcinoma is controversial, with diverse treatment options recommended by different guidelines. There have been no randomized trials that compared outcomes between the various treatment modalities. The PDQ guidelines state that patients with Stage III disease may be treated either by surgery with or without postoperative radiotherapy or by definitive radiotherapy, with surgery reserved for salvage of radiation failures. The State of the Art Oncology, British Columbia, and Peter MacCallum Cancer Institute treatment guidelines recommend radiotherapy (postoperative or definitive) for all patients. Contemporary Australian practice is to preserve the larynx. In randomized trials,62–70 it was shown that larynx preservation with chemoradiotherapy (using laryngectomy for salvage) provided laryngeal preservation in the majority of patients without any detrimental effect on survival compared with laryngectomy. Therefore, it is considered reasonable to offer radiotherapy to all patients with T3 laryngeal carcinoma and to reserve surgery for salvage.

Some guidelines state that a small proportion of patients with early-stage laryngeal carcinoma may undergo larynx-conserving surgery, such as hemilaryngectomy, although most of these guidelines (PDQ, National Comprehensive Cancer Network, and British Columbia Cancer Agency) state that radiotherapy is the preferred treatment. The proportion of patients with supraglottic laryngeal carcinoma who could undergo larynx-preserving surgery in preference to radiotherapy was varied between 0% (i.e., all patients with supraglottic laryngeal carcinoma receive radiotherapy) and 26% (i.e., the proportion of patients with supraglottic laryngeal carcinoma identified as suitable for larynx preservation in a single, large head and neck practice).48

Laser therapy and conservative laryngeal surgery have been described in the literature as treatment options for some patients with early-stage glottic carcinoma. These procedures have not been compared with external beam radiotherapy in a randomized controlled trial, nor have the clinical attributes required to select patients for these procedures been identified. Therefore, it is difficult to estimate the proportion of patients who may undergo these treatments instead of radiotherapy. Because studies have not identified the proportion of patients who may be treated with laser or conservative surgery, we performed sensitivity analysis for 0–10% of patients with early-stage glottic carcinoma being suitable for these treatments.

The indications for radiotherapy (either as definitive treatment or as adjuvant to surgery) in patients with squamous cell carcinoma of unknown primary origin metastatic to the head and neck are controversial. Some groups, such as Maulard et al.,71 advocate routine radiotherapy in patients with localized disease. The main intent is to not only maximize locoregional control in the neck but also to attempt to prevent the occult primary from becoming symptomatic. Others, like Sinnathamby et al.,54 advocate that modern staging techniques obviate the routine use of radiotherapy for unknown primary and propose that radiotherapy should be reserved for patients who are at higher risk of locoregional failure (particularly those with advanced neck disease). O'Mara et al.72 suggest neck dissection as treatment for patients with N1–N2A disease, and Sinnathamby et al. discuss omitting radiotherapy for patients who have solitary lymph nodes measuring up to 6 cm (N1–N2A). The Peter MacCallum Cancer Institute has developed treatment pathways for unknown primary squamous cell carcinoma of the head and neck (unpublished results) in which either preoperative or postoperative radiotherapy is recommended for patients who have > N2A disease, and radiotherapy is considered optional for patients who have N1–N2A disease. In the radiotherapy utilization tree, patients with N1–N2A disease do not receive radiotherapy; this is to avoid a bias toward overestimating the overall optimal radiotherapy utilization rate. Patients with > N2A disease routinely are recommended for radiation. However, in the sensitivity analysis, the branch point underwent modeling, with the proportion of patients receiving postoperative radiotherapy varying between those with ≥ N1 disease and those with ≥ N2A disease.

Sensitivity Analysis

Sensitivity analysis assesses the impact of varying the value of uncertain data on the overall optimal radiotherapy utilization rate. For the head and neck radiotherapy utilization tree, five data items were identified as uncertain: 1) No data were identified to estimate the proportion of patients with early-stage oral cavity carcinoma who should undergo surgery. An arbitrary value of 0.9 was chosen based on reported practice in South Australia, and the sensitivity analysis varied this proportion between 0% patients undergoing surgery (and all patients receiving radiation) and 90% of patients undergoing surgery. 2) The ideal proportion patients with carcinomas of the lip who should undergo surgery alone is uncertain, as discussed above. Therefore, we used incidence data to model between patients who received radiotherapy for tumors that measure > 4 cm in size as the smallest case scenario for radiotherapy (6% incidence) and tumors that measure > 2 cm as the greatest case scenario for radiotherapy (25% incidence). 3) The proportion patients with supraglottic laryngeal carcinoma who could undergo larynx-preserving surgery in preference to radiotherapy was varied between 0% (i.e., all supraglottic laryngeal carcinomas were treated radiotherapeutically) and 26% (i.e., the proportion of supraglottic laryngeal carcinomas identified as suitable for larynx preservation in a single, large head and neck practice). 4) The proportion of patients with early-stage glottic carcinoma who are suitable to undergo conservative surgery was varied between 0% and 10%. 5) The proportion of patients with unknown primary squamous cell carcinoma was varied between patients with lymph node status ≥ N1 receiving radiation (22%) and patients with lymph node status ≥ N2a (9%) receiving radiation.

To assess the impact of these uncertainties on the overall estimate of the need for radiotherapy in all patients with head and neck carcinomas, a univariate sensitivity analysis was performed for each of the variables. This is illustrated as a tornado diagram (see Fig. 2), which is a set of one-way sensitivity analyses brought together in a single graph. In the tornado diagrams, each bar represents a single one-way sensitivity analysis, and the legend provides details of each of the analyses depicted. The variables are ranked by their effect on the overall radiotherapy utilization estimate, with the variables that have most impact appearing at the top of the graph and those with smaller impact appearing below. Figure 2 shows that the proportion of patients with carcinoma of the head and neck who should receive radiotherapy based on evidence and incidence of attributes for radiotherapy is 74% (the dashed line) and within the range of 73–81%, allowing for these uncertainties in sensitivity analysis. In some of the radiotherapy utilization reports that we published previously on other disease sites, a Monte-Carlo analysis was performed in which each of the variables used in the sensitivity analysis were varied simultaneously.7, 9, 11, 13 This was not possible for the head and neck tree, because not all of the sensitivity analyses were for data variation, and many analyses were based on modeling between two treatment alternatives, such as surgery versus radiotherapy.

Figure 2.

Tornado analysis of the variation in data and uncertainty in evidence for radiotherapy in patients with head and neck carcinoma. Blue: proportion of patients with Stage I–II carcinoma of the oral cavity undergoing surgery (0.0–0.9); green: proportion of patients with operable carcinoma of the lip (0.75–0.94); orange: proportion of patients with supraglottic laryngeal carcinoma suitable for preservation surgery (0.0–0.16); gray: proportion of patients with early-stage glottic carcinoma for whom radiotherapy is an alternative; white: proportion of patients with unknown primary head and neck tumors who have lymph node disease that does not warrant routine radiotherapy (0.09–0.10). A description of the interpretation of the tornado diagram is provided in the text (see Sensitivity Analysis).

DISCUSSION

In this study, we found that, optimally, 74% of all patients with head and neck carcinoma (including unknown primary tumors and lip carcinoma) should receive radiotherapy according to evidence-based treatment guidelines for head and neck carcinoma. By tumor site, the recommended optimal radiotherapy rates were oral cavity, 74%; lip, 20%; larynx, 100%; oropharynx, 100%; hypopharynx, 100%; nasopharynx, 100%; paranasal sinus, 100%; salivary gland, 87%; and metastatic squamous cell carcinoma of the neck, 90%. There are several areas within the management of head and neck carcinoma that remain uncertain. However, sensitivity analysis has shown that these uncertainties have little impact on the overall result. All radiotherapy recommendations quoted from evidence-based clinical guidelines were based on Level III and IV evidence, although there is Level II evidence that radical chemoradiation is superior to radiotherapy alone for advanced laryngeal carcinoma.61

It is acknowledged that radiotherapy utilization rates of 100% may be considered unachievable. However, there are some tumor subsites in the head and neck where radiotherapy is the mainstay of treatment for all stages of disease (e.g., nasopharyngeal carcinoma), because it is superior in terms of outcome or morbidity compared with the alternative treatments of chemotherapy alone or surgery. In other tumor subsites, the optimal radiotherapy utilization rate of 100% does not necessarily mean that all patients are suitable for radical radiotherapy. However, even in patients with metastatic disease and relatively poor performance status, it is usual for patients to have symptoms attributable to the primary or regional disease that would justify a short course of palliative radiotherapy. Rare instances of patient refusal to undergo any form of therapy or when radiotherapy truly is not recommended are likely to represent < 1% of patients73 and, hence, are unlikely to have a significant impact on the overall optimal utilization rate of 100% for that subsite. Actual radiotherapy utilization rates from Sweden in 20004 have shown that utilization rates of 100%, in fact, are achievable in clinical practice.

Comparison with Actual Practice

The optimal rates for each head and neck tumor site were compared with international data from the Surveillance, Epidemiology, and End Results (SEER) data base,73 the South Australian Cancer Registry,41 the Swedish Radiotherapy Utilization Survey,4 and the Yorkshire Cancer Registry.74 Two periods were examined in the SEER data base. The periods chosen were the most recent for which data were available (1995–2000) and 20 years prior (1975–1980) to evaluate changes in actual utilization within a single study population over time. Table 3 shows that radiotherapy utilization rates in South Australia, the United States, and Yorkshire were lower than optimal rates. There were some data to suggest an increased use of radiotherapy over time according to the SEER data (e.g., in salivary gland carcinoma, there was an ≈ 16% increase in use over 20 years). However, there also were decreases in the use of radiotherapy for nasopharyngeal carcinoma and hypopharyngeal carcinoma over the past 20 years. The reasons for these reductions in use could not be identified.

Table 3. Comparison of Optimal and Actual Radiotherapy Utilization Rates
Head and neck tumor subsiteOptimal RT utilization rates (%)Actual RT utilization rates (%)
Sweden, 2002 (Moller et al., 20034)SA State Cancer Registry (Australia) 1990–1994 (Luke et al., 200341)North Yorkshire (UK) 1999 (NYCRIS, 199974)SEER (US) (NCI, 200273)
1975 to 19801995 to 2000
  • NYCRIS: Northern and Yorkshire Cancer Registry and Information Service; SEER: Surveillance, Epidemiology, and End Results; NCI: National Cancer Institute; NR: not reported; NRS: not reported separately.

  • a

    Includes brachytherapy.

  • b

    Data include the rate for sites in the salivary gland.

  • c

    Data were included with sites in the oral cavity.

Oral cavity7494a44b56  
Lip20222NR67
Larynx10010080726574
Pharyngeal tumors100NR81NRNRNR
Oropharynx100100NRSNR7669
Hypopharynx10039NRSNR7473
Nasopharynx100100NRSNR8580
Paranasal sinus10010055NRNRNR
Salivary gland8760cNR3854
Metastatic unknown primary90NRNRNRNRNR

The data on actual utilization of radiotherapy show that some tumor subsites have a reasonable correlation between actual and optimal radiotherapy rates (e.g., actual radiotherapy utilization rates in Sweden are very close to the optimal rates for carcinomas of the nasopharynx, hypopharynx, larynx, and paranasal sinuses). However, significant shortfalls also were identified for some populations. There are large differences between the SEER actual radiotherapy utilization rates versus the optimal rates for salivary gland tumors and tumors of the hypopharynx. Reasons for these differences are not available. Some differences may be due to differences in stage distribution and histologic subtype distribution in differing populations, but it is unlikely that the differences solely relate to differing disease presentation. It is possible that real shortfalls exist in radiotherapy treatment delivery, and this may reflect differences in opinion regarding the role of radiotherapy. This underlines the need for good-quality, high-level (Level I–II) evidence for the role of radiotherapy. Unfortunately, most of the evidence is of poorer quality.

Actual rates of head and neck radiotherapy from Sweden in 20004 showed that actual and optimal rates were identical for the types of head and neck tumors for which 100% of tumors should receive radiation (larynx, nasopharynx, oropharynx, paranasal sinuses) and showed that, for some tumor subsites, a 100% rate is achievable. For patients with oral cavity tumors in Sweden, the actual rate of utilization was higher than the optimal rate, although the actual rate included brachytherapy, which appears to be more prevalent than surgery in the management of oral cavity carcinomas in Sweden. We may have underestimated the optimal radiotherapy utilization rate for oral cavity tumors because of a lack of evidence for the superiority of either surgery or radiotherapy as primary treatment.

A survey of radiotherapy departments in New South Wales (NSW), Australia,75 assessed the total number of patients with head and neck carcinoma (excluding lip carcinoma and unknown primary tumors) who received radiotherapy in 2000. The data were not separated by subsite location, and the actual radiotherapy utilization rate reported was 86%. The optimal rate for the same cohort of patients with head and neck carcinoma (i.e. excluding lip carcinoma and unknown primary tumors), as calculated by using the head and neck radiotherapy utilization tree, was 89%. This close correlation between actual and optimal radiotherapy utilization for patients with head and neck carcinoma in NSW is in stark contrast to most other tumor sites, for which we have reported larger discrepancies between actual rates and optimal rates.7, 9, 11, 13–15 Possible reasons why the actual rates are similar to the optimal rates in head and neck carcinoma may include that patients with head and neck carcinoma in NSW generally are treated in specialist units (as opposed to other types of carcinoma, for which care is distributed more widely among clinicians), the high use and proven efficacy of radiotherapy in head and neck carcinoma, and the existence of well accepted evidence on the use of radiotherapy for head and neck carcinoma, even though some of the evidence is not randomized. Further study into the reasons behind the convergence of optimal and actual radiotherapy utilization rates in head and neck carcinoma may help to improve the gap between actual and optimal radiotherapy utilization rates for other tumor sites.

In conclusion, for this study, we used an evidence-based method to estimate the optimal radiotherapy utilization rate for head and neck malignancies. The recommended proportion of all patients with head and neck carcinoma who, according to the best available evidence, should receive at least 1 course of radiotherapy is 74%. This optimal utilization rate is substantially higher than current practice in South Australia and the United States, although it is similar to current practice in Sweden and to the practice in 2000 in NSW. We believe that further epidemiological patterns-of-care studies on the use of radiotherapy for patients with head and neck malignancies are required.

Acknowledgements

The authors thank Professor L. Peters, Dr. D. Dalley, Dr. K. Clark, Dr. M. Veness, Dr. R. Smee, and the members of the Steering Committee of the Australian National Cancer Control Initiative for comments concerning the study design and radiotherapy utilization trees.

Ancillary