Temporal Trends in the Treatment of Early- and Advanced-Stage Laryngeal Cancer in the United States, 1985-2007

Objective To describe trends and 4-year survival rate of surgical and nonsurgical treatment for laryngeal cancer.

Design Observational cross-sectional study.

Patients A total of 131 694 cases of laryngeal cancer diagnosed from 1985 to 2007 identified from the National Cancer Database.

Main Outcome Measures Primary treatment information, including radiation therapy (RT), chemoradiation (CRT), and curative intent surgery, were identified. The association between treatment and the patient's clinical and nonclinical variables was analyzed using univariate and multivariate statistics. The 4-year survival rate was generated through Kaplan-Meier estimates, and multivariate Cox proportional hazard models were used to generate hazard ratios.

Results Among patients with early-stage cancer, the proportion receiving primary surgery increased (from 20% in 1985 to 33% in 2007), whereas the use of RT decreased from 64% to 52%. Patients with early-stage cancer who resided in areas with higher socioeconomic status (SES) zip codes, had private insurance, who were not African American, and who were treated at academic facilities were more likely to receive surgery. The 4-year survival rate for patients with early-stage laryngeal cancer treated with surgery was higher than the rate for those treated with RT (79% vs 71%). Among patients with advanced-stage cancer, the use of CRT increased from less than 7% to 45%, whereas the use of total laryngectomy decreased from 42% to 32%. The use of CRT was more common among patients who resided in areas with higher SES zip codes, had private insurance, and who were younger. The 4-year survival rates for patients with advanced laryngeal cancer treated with total laryngectomy, CRT, and RT were 51%, 48%, and 38%, respectively. Factors associated with an increased risk of death from advanced laryngeal cancer included receiving CRT and race/ethnicity.

Conclusions Among patients with early-stage laryngeal cancer, we observed an increasing proportion of primary surgical therapy during this study period. Among patients with advanced-stage cancer, we observed an increasing proportion of CRT. Not only were clinical factors associated with type of treatment, but select sociodemographic elements were also associated with treatment. Further investigation as to the decision-making process of patients with different sociodemographic backgrounds will assist in mitigating the differences in survival for this group of patients.

Laryngeal cancer was diagnosed in nearly 13 000 individuals in the United States¹ and 3660 of them were projected to die from this disease in 2010. Early-stage cancer has traditionally been treated with radiation therapy (RT) or surgical resection with comparable successful outcomes.²Quiz Ref IDPrior to the Veterans Administration (VA) Laryngeal Cancer Study, which was published in 1991,³ treatment for advanced laryngeal cancer was total laryngectomy; however, this landmark study established the usefulness of chemoradiation (CRT) as a valid treatment option for advanced laryngeal cancer. More than 300 patients with stage III or IV laryngeal cancer were randomized to 2 arms of treatment: induction chemotherapy with cisplatin and fluorouracil and radiotherapy, or total laryngectomy followed by RT. Sixty-two percent of patients in the CRT arm were able to preserve their larynx. Furthermore, no significant differences in overall survival were evident between the nonsurgical and surgical arms of the study. A subsequent trial by the Radiation Therapy Oncology Group (RTOG 91-11)⁴ compared 3 different types of nonsurgical treatment (RT alone, induction chemotherapy followed by RT, and concurrent CRT) for advanced laryngeal cancer. Concurrent CRT was reported to offer a survival rate comparable with that of larygnectomy.⁴ This 2003 study solidified the use of CRT as a valid treatment option for patients with advanced laryngeal cancer.

The purpose of this study is to investigate how patterns of treatment for both early- and advanced-stage laryngeal cancer have changed from 1985 to 2007 using data collected by the National Cancer Database (NCDB). The objectives of the study are as follows:

1. To describe temporal trends of surgical and nonsurgical treatment for early- and advanced-stage laryngeal cancer.

2. To describe clinical and nonclinical factors, including treating facility type, associated with type of treatment.

3. To report the 4-year survival rates for early- and advanced-stage laryngeal cancer by treatment.

Data from the NCDB, a hospital-based cancer registry jointly sponsored by the American Cancer Society and the American College of Surgeons, were used in this study. The NCDB includes approximately 70% of all malignant cancers in the United States from over 1400 facilities accredited by the American College of Surgeons' Commission on Cancer (CoC) that collect and submit data to the NCDB.⁵ The NCDB contains standardized data elements on patient demographics, insurance status, stage at diagnosis, and zip code–level socioeconomic factors and facility-level factors. Data are entered using the standardized data dictionary of the NCDB: the Facility Operations and Data Standards Manual.⁶ The NCDB requires hospital registries to update vital status and other information in 5-year cycles; for example, patients first diagnosed in 1998 would be initially reported in 2000 and have their vital status updated in 2005 (in the same “call for data” cycle as the 2003 incident cases). The Morehouse University (Atlanta, Georgia) institutional review board (IRB) reviewed this study and determined that the study is exempt from IRB review.

Patients diagnosed from 1985 to 2007 with a first primary invasive squamous cell carcinoma of the larynx were selected (179 551 individuals). Patients younger than 18 years or older than 99 years were excluded (90 individuals). Although we selected only invasive tumors, data from 393 patients who were miscoded as having stage 0 tumors were removed from the analyses. Patients with evidence of clinical distant metastasis (2702 individuals); who did not receive treatment (10 594) or who were missing treatment (18 960), facility type (519), or sex (38); or who were recorded as having received photodynamic therapy or electrocautery surgery (24) were also removed from the analyses. Owing to insufficient numbers, data from patients with other forms of government insurance (which includes insurance through the Bureau of Indian Affairs and the military [893 patients]) were also removed from the analyses. To minimize the impact of changes in CoC accreditation over time on the outcome of our study, we removed data from patients who were not from continuously accredited CoC programs from further analysis (13 914). The total analytic cohort included 131 694 patients.

Patients were grouped into the following 5 treatment types: chemotherapy with concurrent radiation (CRT), which was defined as having received chemotherapy and radiation within 90 days of one another; RT alone; chemotherapy alone; invasive surgery (which included total and subtotal laryngectomy); and local surgery. Local surgery included all endoscopic resections, such as laser excision, stripping, and excisional biopsy. Although our definition of CRT included patients receiving chemotherapy and radiation within 90 days of one another, most patients (76%) received concurrent chemotherapy and RT, defined as occurring within 42 days (6 weeks) of one another. Patient-level independent variables included age at diagnosis (18-49 years, 50-59 years, 60-69 years, and 70-99 years); sex; diagnosis year; and race/ethnicity, which was coded as non-Hispanic white, Hispanic, African American, other/Asian, and missing race/ethnicity. If clinical stage was missing, pathologic stage was relied on. Early-stage cancer was defined as American Joint Committee on Cancer (AJCC) TNM stage I or II, and advanced-stage was defined as stage III or IV. Cancer subsite was also considered and categorized into 3 groups: (1) supraglottic, (2) glottic, and (3) subglottic.

Because of the small proportion of subglottic tumors, we excluded these tumors as well as patients with missing data on cancer subsite from multivariate analysis. Since 1996, the NCDB has collected insurance status, which enabled us to examine the impact of insurance status on treatment in more recent years (1998-2007). We did not include the first 2 years (1996 and 1997) of insurance data to allow time for more complete and accurate coding. Primary payers were grouped into the following: Medicaid, uninsured (which includes not insured, charity write-off, and self-pay), private insurance plans (health maintenance organizations, preferred provider organizations), and Medicare. Because Medicare is available to most patients 65 years or older but only for permanently disabled individuals younger than 65 years, we created a separate category for younger (18-64 years) Medicare patients. An area-based indicator of socioeconomic status (SES), the proportion of the population that did not have a high school diploma, was derived at the zip code level from 2000 US Census Bureau data⁷ and included as quartiles of the observed distribution in the general US population for patients diagnosed from 1996 through 2007. For patients diagnosed from 1985 through 1995, the 1990 US Census Bureau data⁸ were used to estimate area-level education. Comorbidity data in the NCDB are available only for diagnosis years 2003 and after and are collected from the hospital face sheet. Therefore, there is more potential bias for reporting higher comorbidity for inpatient cases (ie, major surgery) than for outpatient cases, since we do not have the medical claims history to verify comorbidity. For these reasons, we did not include comorbidity in this analysis.

Based on the CoC approvals classification, we categorized facilities into the following groups: community facilities, community cancer centers, and teaching or research centers. Community facilities treat at least 300 patients with cancer per year and have a full range of services for cancer care, but patients need a referral for portions of their treatment. Community cancer centers are facilities that offer the same range of services as the community facilities but treat at least 750 patients with cancer annually and conduct weekly cancer conferences. Teaching or research facilities have residency programs and ongoing cancer research. Patients with missing treatment facility type were included in a separate category.

Statistical analyses were performed with SAS software (version 9.2; SAS Institute Inc, Cary, North Carolina). χ² Tests and the corresponding 2-tailed P values were calculated to analyze the relationship between facility type and all other covariates. Multivariate logistic regression models were used to estimate odds ratios (ORs) and 95% CIs of select treatments for patients diagnosed from 1998 through 2007 among early- and advanced-stage tumors separately. Factors associated with local surgery vs RT were examined. Among patients with advanced-stage cancer, factors related to CRT vs total laryngectomy were examined. We also conducted multilevel models to account for zip code–level clustering.⁹ All-cause survival Kaplan-Meier estimates were generated, and adjusted hazard ratios (HRs) and 95% CIs were estimated using Cox proportional hazard models for early- and advanced-stage cancers separately. Despite the availability of the 5-year survival rate for the most recent diagnosis years (1998-2002), we examined the 4-year survival rate to ensure a lower loss to follow-up rate, which was 14% in our analytic cohort. The NCDB does not collect the cause of death; therefore, we used observed all-cause mortality in our study. Among the cohort with early-stage cancer, only patients receiving local surgery or RT were included in the survival analysis because patients receiving other forms (total laryngectomy, CRT, or chemotherapy alone) of treatment were not considered to have complete treatment or the treatment was generally not recommended. Among the cohort with advanced-stage cancer, only patients receiving laryngectomy or CRT were considered for multivariate survival analyses.

Data for a total of 131 694 patients were analyzed. Most of the patients were male (79.0%), were older than 60 years (64.1%), and were treated at a community cancer center or teaching-research center (86.0%). Slightly more than half of all patients (52.5%) were diagnosed as having early-stage cancer (stages I and II). Radiation therapy was the most common primary form of treatment for early-stage cancers (41.6% of patients), whereas laryngectomy and CRT were the most common forms of treatment among patients with advanced-stage cancers (Table 1 and Table 2). The proportion of advanced-stage cancers increased from 35.0% in 1985 to 47.0% in 2007 (Figure 1); however, the rate of advanced-stage cancers remained relatively stable over the more recent time period (1998-2007) considered in multivariate logistic models predicting treatment.

A multivariate logistic regression analysis examined the effect of race/ethnicity, age, year of diagnosis, SES, and type of hospital on different primary treatments for early-stage laryngeal cancer (Table 2). Patients who were African American (OR, 0.81; 95% CI, 0.74-0.89), uninsured (OR, 0.78; 95% CI, 0.68-0.90), Medicaid recipients (OR, 0.79; 95% CI, 0.70-0.90), and residing in zip codes with less than 14.0% of the population not receiving high school diplomas (OR, 1.31; 95% CI, 1.22-1.41) were statistically significant more likely to receive surgery. In multilevel models adjusting for clustering among patients residing in the same zip codes, the effect of neighborhood SES on treatment for patients with early-stage cancer remained statistically significant (data not shown). In addition, patients treated at community cancer centers (OR, 0.81; 95% CI, 0.76-0.87) and teaching research facilities (OR, 0.89; 95% CI, 0.82-0.96) were also less likely to receive surgery based on multivariate analyses. Trends in treatment for patients with early-stage laryngeal cancer show slightly decreasing rates for RT and a slowly increasing role for surgical excision until approximately 1995, when rates subsequently plateaued (Figure 2A).

A multivariate logistic regression analysis was also used to examine the effect of race/ethnicity, age, year of diagnosis, SES, and type of hospital on different primary treatments for advanced-stage laryngeal cancer (Table 3). Medicaid recipients (OR, 0.88; 95% CI, 0.81-0.97) were less likely to receive CRT. Patients living in neighborhoods with higher proportions of high school graduates were more likely to receive CRT than laryngectomy. In multilevel models adjusting for clustering among patients residing in the same zip codes, the effect of neighborhood SES on treatment for patients with advanced-stage cancer was still statistically significant (data not shown). In addition, individuals treated at teaching-research facilities (OR, 0.70; 95% CI, 0.62-0.76) were less likely to receive CRT. African American patients were more likely (OR, 1.14; 95% CI, 1.04-1.24) to receive CRT than laryngectomy. Trends in treatment for advanced laryngeal cancer demonstrate rapid increase of CRT. Beginning in 1995, CRT increased from 10% to 46% in 2007, surpassing laryngectomy in 2004 as the most common type of treatment for advanced laryngeal cancer (Figure 2B).

The 4-year survival rates among patients with early-stage cancer receiving local surgery and radiation were 79% and 71%, respectively. Among patients with advanced-stage cancer, the survival rates for patients receiving laryngectomy, CRT, and RT alone were 51%, 48%, and 38%, respectively. Patients with early-stage cancer receiving local surgery had a lower hazard ratio of death (Table 4), and African American patients had a higher risk of death. Patients with advanced-stage cancer receiving CRT had higher risks of death compared with patients treated with laryngectomy, and African American patients also had a higher hazard of death after 4 years of follow-up (Table 5). For both early- and advanced-stage cancers, patients treated at teaching-research facilities had a lower HR compared with those treated at community facilities.

The use of RT and surgery has remained the primary form of treatment for patients with early-stage laryngeal cancer. Quiz Ref IDDuring our study period, surgical resection of early-stage laryngeal cancer increased with a corresponding decrease in radiation. Advances in transoral endoscopic resection during this time may explain the rise in the use of surgery for these early-stage cancers^2,10 The advantages of surgery over RT include less expense, shorter treatment duration (1-2 days vs 6 weeks of RT), and the ability to be repeated. How a patient decides among competing treatments is not well represented in the NCDB or in any database. Thus, the differences in treatment can be measured to some degree by the variables that are available in the data set. However, cultural biases in treatment may not be reflected in such a database. In 2002, the Cochrane Database of Systematic Reviews¹¹ found insufficient evidence in the literature to make a global determination as to which was the ideal treatment for early-stage cancer. More recently, 1 study found that patients with early-stage laryngeal cancer, independent of treatment, treated at a high-volume institution had a survival benefit higher than that of patients seen at a low-volume institution.¹² In this article, high-volume institutions included teaching and academic facilities, and low-volume facilities included community hospitals, as designated by the CoC. Thus, our current study's findings are consistent with those of previous studies.

The 1991 Veterans Affairs Laryngeal Cancer Group demonstrated that survival after nonsurgical therapy is comparable with survival after total laryngectomy.³ A comparable study in Europe reported similar overall survival and a 42% laryngeal preservation rate.¹³ A subsequent study, RTOG 91-11, was designed to determine the best sequencing of nonsurgical treatment. RTOG 91-11 demonstrated that concomitant CRT achieved higher laryngeal preservation rates than induction chemotherapy followed by RT or RT alone without the use of chemotherapy, without a detriment in survival.⁴ Since then, there has been a steady increase in the use of CRT for the treatment of advanced laryngeal cancer by community hospitals, community cancer centers, and teaching-research facilities.^12,14,15Quiz Ref IDOur current study demonstrates that total laryngectomy for advanced laryngeal cancer has declined with a simultaneous increase in the use of CRT, and nonclinical factors were associated with type of treatment.

Our study also examined if nonclinical factors affected the treatment of early- and advanced-stage laryngeal cancer. We observed lower odds of local surgery among African Americans with early-stage disease and higher odds of CRT among African Americans with advanced-stage disease. Other authors, such as Shavers et al,¹⁶ have also reported on the association between demographic factors and type of treatment. Shavers et al¹⁶ observed that African Americans were significantly less likely than whites to receive cancer-directed surgery for regional and distant stage cancers of the oral cavity and pharynx. Molina et al¹⁷ published a study on patients with head and neck cancer in Florida from 1998 to 2002 using a linked cancer registry and claims data set. They also reported that fewer African American patients underwent surgery than did whites (32% vs 45%; P < .001). Our observation of treatment variations by race/ethnicity may be indicative of cultural attitudes toward cancer treatment. For example, among Latina patients with breast cancer, family influence seemed to have a profound effect on treatment decision making.^18,19 Chinese women were found to prefer mastectomy over breast-conserving treatment because of their cultural belief that a mastectomy was safer and their feeling pressure to conform to this belief.²⁰ These studies underscore the importance of communication between physician and patient in a culturally sensitive and respectful way.

Insurance status was also found to be associated with type of treatment in this study. Other studies have also reported an association between insurance and type of treatment for cancers other than head and neck cancer.^21-24 For example, a 1988 study among lung cancer patients by Greenberg et al²¹ discovered that patients with lung cancer and private insurance were 52% more likely to receive surgery than those without private insurance. The preference for surgery over other treatment modalities for those with private insurance is not commonly reported. In an article specifically aimed at head and neck cancer, Kwok et al²⁵ reported that patients diagnosed as having head and neck cancer and who had Medicaid, Medicare, or no insurance had increased risk for death in their single institution. Furthermore, this result was independent of variables such as cancer stage, treatment, age, race, smoking, and alcohol use. However, that finding was not the result of a multivariate analysis and thus could not control for many critical factors, including cancer stage, cancer location (oral, pharynx, or larynx), and race. Quiz Ref IDOur finding that Medicaid and uninsured patients receive surgery more often than those with private insurance, among patients with advanced-stage cancer, is consistent with these findings if we consider Medicaid and uninsured insurance statuses as a proxy for low SES. It has been reported that patients with low SES have diminished access to health care, and prolonged follow-up may be not convenient or feasible.²⁶ Indeed, in our study, we report that individuals living in zip codes with a higher percentage of high school graduates were more likely to receive local surgery over RT for early-stage disease. However, among patients with advanced-stage cancers living in zip codes with higher SES measures were more likely to receive CRT over laryngectomy, even after adjusting for zip code–level clustering.

Our study demonstrates that treatment of advanced-stage cancer has changed significantly in the past decade. In the most recent year of our data (2007), half of all patients with advanced-stage cancer were treated with CRT, and less than one-third were treated with laryngectomy. The increasing reliance on CRT, first noticed in 2006, is not unique to laryngeal cancer and has been seen for other advanced head and neck cancers as well.²⁷ For the patient, choosing between CRT and laryngectomy is not a trivial decision. Laryngectomy may result in a total loss of voice and sometimes impairment of swallowing function, leading to a decreased quality of life.²⁸ However, treating advanced laryngeal cancer with CRT can result in complications such as renal failure and bone marrow suppression. Later complications, such as persistent dysphagia, gastrostomy tube dependency, pharyngoesophageal stenosis, chronic lung aspiration, and permanent tracheotomy dependence, can occur.²⁹ In addition, a proportion of patients receiving CRT may still need a laryngectomy because of a dysfunctional larynx and/or persistent disease.⁴ One study reported a 7% salvage laryngectomy rate over a 10-year period but with a 20% fistula rate.³⁰

Hoffman et al¹⁴ reported in 2006 that survival for laryngeal cancer was declining. This relationship has been replicated in subsequent studies.^27,31 These analyses seem to contradict previous reports from clinical trials that found an equivalent survival rate for those treated with CRT and laryngectomy, possibly owing to differences in the patient population between those enrolled in clinical trials and patients who are treated at CoC facilities across the United States.³ In addition, a recent study reports an increased risk of death with CRT compared with laryngectomy.²⁷ Previous studies have also reported that treating facility type and volume were associated with survival,³¹ and treatment at high-volume teaching-research facilities was significantly associated with improved survival, after controlling for clinical and nonclinical characteristics. Quiz Ref IDIn the current analysis, we again demonstrate a statistically significant decreased risk of death for patients with both early- and advanced laryngeal cancer treated at teaching-research facilities.

A limitation of this analysis on the relationship between nonclinical factors and treatment of laryngeal cancer is the lack of information on the treatment options that were offered to the patient or the reasoning behind those recommendations. Numerous nuances may influence a patient's final choice of treatment, including functional status of the larynx, comorbidities, and patient preferences. First, the NCDB also does not collect information on complications due to treatment. Second, the reporting of a patient's insurance status is not flawless. Patients categorized as Medicaid insured often initially present with no insurance coverage and are retroactively enrolled in Medicaid at the time of diagnosis, so there is possible underreporting of uninsured patients in the NCDB. In addition, health insurance is not a static variable for patients, and the lack of insurance history for patients makes it impossible to determine whether the choice of treatment was solely affected by a patient's insurance status at the time of diagnosis.³² Third, we did not have individual data on income, education, or other socioeconomic variables and had to estimate based on the demographics of the patient's zip code. Fourth, because our NCDB data come only from CoC-approved hospitals, the patients and treatment patterns may not be fully representative of all patients with laryngeal cancer in the United States. However, the NCDB does cover approximately 70% of all incident malignant cancer cases in the United States. Fifth, NCDB does not collect cause of death.

While much controversy exists over the current treatments of laryngeal cancer, it is important to recognize that several new modalities are on the horizon and will challenge our accepted gold standards of RT, CRT, and surgery. Photodynamic therapy has been shown to cure patients with early-stage oral and laryngeal dysplasia in several studies, and patients have experienced fewer long-term adverse effects, such as impaired swallowing and speech difficulties.^33-35 Robotic surgery, first used in cardiology and urology, has been shown to decrease operative times required for head and neck surgical procedures, as well as decrease blood loss and shorten hospital stays.^36,37 Further studies are necessary and are under way to determine whether these new technologies are an improvement on existing practice guidelines.

In conclusion, to our knowledge, this analysis provides the first assessment on how several factors, such as race and insurance status, predicate which treatment a patient with early- or advanced-stage laryngeal cancer receives, independent of numerous clinical and nonclinical variables among a large national cohort. Additional studies are necessary to determine why these biases exist and if they are clinically significant. While the use of RT and surgery for early-stage laryngeal cancer has been consistent over the past 2 decades, treatment of advanced laryngeal cancer has shifted to the extensive use of CRT with a concomitant drop in TL and survival. We recognize that the selection of treatment for patients with laryngeal cancer is a complex process and urge great discretion when choosing a treatment.

Correspondence: Amy Y. Chen, MD, MPH, American Cancer Society, National Home Office, 250 Williams St NE, Atlanta, GA 30303 (achen@emory.edu).

Submitted for Publication: May 27, 2011; final revision received June 24, 2011; accepted August 9, 2011.

Author Contributions: All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Chen and Fedewa. Acquisition of data: Chen and Fedewa. Analysis and interpretation of data: Chen, Fedewa, and Zhu. Drafting of the manuscript: Chen, Fedewa, and Zhu. Critical revision of the manuscript for important intellectual content: Chen and Fedewa. Statistical analysis: Fedewa. Administrative, technical, and material support: Chen, Fedewa, and Zhu. Study supervision: Chen.