Drs. Ang, Chen, Curran, Garden, Harari, Murphy and Wong are founding and current members of the LORHAN advisory board.
Detailed information about how patients with head and neck carcinoma (HNC) are treated across practice settings does not exist. The authors conducted a prospective, observational study to examine the patterns of care for a series of patients with newly diagnosed HNC in the United States and to test 2 hypotheses: 1) There is no difference in the pattern of care between community and academic settings; and 2) the results of major randomized clinical trials will change the pattern of care in both practice settings within 1 year of publication in peer-reviewed journals.
Patients aged ≥18 years were enrolled in the Longitudinal Oncology Registry of Head and Neck Carcinoma (LORHAN) after providing written informed consent if they had a confirmed diagnosis of new HNC and were scheduled to receive treatment other than surgery alone.
Between 2005 and 2010, 100 centers enrolled 4243 patients, including 2612 patients (62%) from academic investigators and 1631 patients (38%) from community centers. Initial treatments were radiation with concurrent chemotherapy (30%) or cetuximab (9%), adjuvant radiotherapy (21%), induction chemotherapy (16%), and other (24%). Intensity modulated radiation therapy was the dominant radiation technique (84%). Single-agent cisplatin was prescribed in nearly half of patients and more often in academic centers (53% vs 43% of patients; P < .0001). Single-agent cetuximab was the next most common drug used (19%) and was prescribed more frequently in community settings (24% vs 17%; P = .0001). The data rejected the 2 prospective hypotheses.
Approximately 49,000 new cases of head and neck cancer (HNC) are diagnosed each year in the United States.1 Most patients with HNC present with locally advanced disease,2 diminishing their chance for a cure. Several clinical trials3-7 and meta-analyses8, 9 established that adding cisplatin concurrently to radiation improved the rates of locoregional control, organ preservation, and/or survival for patients with HNC. Four recently reported randomized trials established new therapy options. Bonner et al reported on a trial of cetuximab plus radiation versus radiation alone in patients with locally advanced HNC10 and demonstrated that adding cetuximab to radiation significantly improved locoregional control and overall survival (OS) without increasing the grade ≥3 toxicities commonly associated with radiation, including mucositis. Two phase 3 trials addressed induction chemotherapy with docetaxel plus cisplatin and 5-fluorouracil (TPF) versus cisplatin and 5-fluorouracil (PF) alone followed by radiation or chemoradiotherapy11, 12 and demonstrated that TPF yielded significantly longer OS compared with PF alone. Rates of grade 3/4 neutropenia and febrile neutropenia were higher for the TPF group in both studies, although treatment delays because of neutropenia, grade 3/4 thrombocytopenia, and fatigue were more frequent in the PF group.11 Pointreau et al also reported that TPF was significantly better than PF alone in preserving the larynx at 3 years in patients with advanced cancers of the larynx and hypopharynx who required total laryngectomy.13 Patients in the TPF group had more grade 4 or febrile neutropenia compared with patients in the PF group, who had more grade 3/4 stomatitis and grade 4 creatinine elevation. It is unclear how quickly and the extent to which these positive trials have changed daily clinical practice and whether and how treatment strategies have varied by practice setting.
The Longitudinal Oncology Registry of Head and Neck Carcinoma (LORHAN) is a national registry that was launched in 2005 with the objective of collecting data on patterns of care for patients with HNC. LORHAN was designed to extend and complement general cancer registries by capturing more details on primary cancer treatments, toxicities, supportive care, and outcomes across practice settings. LORHAN also differs from other registries in that it prospectively identified 2 hypotheses to test:
There is no difference in the pattern of care between community and academic settings for patients with HNC; and
The results of major randomized clinical trials will change the pattern of care in community and academic settings within 1 year of publication in peer-reviewed journals.
Here, we report the first comprehensive analysis of LORHAN.
MATERIALS AND METHODS
A multidisciplinary advisory board led LORHAN and guided the design, conduct, analysis, and reporting. Funding of the registry originally was provided by ImClone Systems Inc. (Branchburg, NJ) and was assumed by Eli Lilly and Company (Indianapolis, Ind) after acquisition of ImClone in 2008. MedNet Solutions, Inc. (Minnetonka, Minn) administered LORHAN, including providing the electronic data capture system, site, and statistical support, and advisory board management. Any physician who was treating HNC and was willing to follow patients for at least 2 years was eligible to participate in LORHAN. Sites were recruited through mass mailings using society lists, the registry website, gatherings at medical conferences, and in-person visits. Investigators received a modest stipend per patient, which was spread out over 3 years to encourage entry of follow-up data. MedNet Solutions Inc. provided site training using an online, interactive tutorial or individual personal instruction. Participating investigators could view their data against the blinded aggregate database at any time. The LORHAN advisory board had full access to data listings and collaborated with MedNet Solutions Inc. on the interpretation and publication of data. This study was registered with ClinicalTrials.gov (NCT01080313).
After institutional review board approval was obtained, eligible patients who provided written informed consent were enrolled. Eligibility criteria were aged ≥18 years and a histologically or cytologically confirmed diagnosis of new carcinoma involving the oral cavity, oropharynx, nasopharynx, hypopharynx, larynx, or neck lymph node metastasis from unknown origin. Patients must have been scheduled to receive treatment other than surgery alone. Principal investigators at each center designated their practice setting as academic or community.
Information collected for LORHAN was considered part of routine care. No additional assessments or evaluations were required for registry participation. Data collected at baseline included sex, race, weight, histology and location of primary tumor, stage of disease and performance status, type of prior surgery (debulking, staging, curative) if applicable, and enrollment in a clinical trial. During treatment, details on radiotherapy and systemic therapies captured included dose, schedule, and duration. Data also were entered on supportive care received, including use of opioid analgesics, antiemetics, and feeding tubes. At the completion of therapy, the best response to treatment and incidence of selected toxicities were recorded. Retreatment information was collected for progressive disease. Patients were followed for up to 4 years for disease and vital status. Outcome data were collected at the end of treatment/retreatment and during follow-up periods.
Data Entry and Management
Data were entered by the physicians or designees into electronic forms through the registry's website. There were up to 4 forms for any given patient: baseline, initial treatment (cancer therapy and supportive care), follow-up, and, when applicable, retreatment. Normal ranges were predefined by the LORHAN advisory board from which hard edit checks were created. Data were subject to manual and automated error checking during electronic entry with procedures that searched for logical inconsistencies, out of range values, and missing data. Error messages were generated when missing or inconsistent data were identified. The form was “locked” once any error messages were addressed and the physician reviewed and digitally signed the form.
MedNet Solutions Inc. enforced restricted access control mechanisms and incorporated encrypted point-to-point data transfer by secure HTTP protocols. Patient and physician confidentiality was strictly maintained.
Study Design and Methods
The primary objective was to describe, in detail, patterns of care for patients with HNC. Secondary objectives were to determine the incidence and severity of major treatment toxicities; to identify supportive care received for managing nutrition, pain, nausea, and other complications, and, subsequently, to document outcomes (tumor control, survival).
The current analysis focused on describing patient characteristics, treatment patterns, and early treatment outcomes. For categorical and ordinal variables, frequencies and percentages were calculated. For continuous variables, descriptive statistics (number, mean, median, standard deviation, and range) were used. To test the null hypothesis, we used chi-square tests, t tests, and other nonparametric tests as required. Survival-based analyses were performed using Kaplan-Meier methodology with censoring as appropriate and were evaluated using log-rank tests with a 2-tailed P value ≤ .05 to reject the null hypothesis. Multivariable logistic regression was used to assess the relation of prespecified independent variables to selected site-related and treatment-related dichotomous measures. Wald chi-square tests and likelihood ratio tests were used to assess the significance of independent variables. Nonsignificant variables (eg, P > .05) were not removed from the final models. Odds ratios and associated 95% confidence intervals also were calculated. All analyses were performed using the statistical software package R (version 2.12.0 or later; R Foundation for Statistical Computing, Vienna Austria; available at: http://www.r-project.org/; accessed April 19, 2012).
In total, 193 investigators from 100 sites enrolled patients. Sites and distribution of patients mirrored data from the Surveillance, Epidemiology, and End Results database (available at: http://statecancerprofiles.cancer.gov/index.html; accessed April 19, 2012) (Fig. 1). The majority (77%) of principal investigators designated their setting as a community practice.
Between December 2005 and September 2010, 4243 patients were enrolled (Fig. 2), including 2612 patients (62%) from academic investigators and 1631 patients (38%) from community centers. Patients who were treated in academic centers were slightly younger but had relatively more advanced disease and poorer performance status. In addition, more African Americans and higher proportions of patients with oropharyngeal and sinonasal tumors were treated in academic centers (Table 1). The distribution of primary sites was similar to data from the North American Association of Central Cancer Registries1 indicating that 78% of new cases originated from the oral cavity and pharynx, and 22% originated from the larynx.
Table 1. Demographic and Clinical Characteristics
No. of Patients (%)
All, N = 3599
Academic, n = 2297
Community, n = 1302
Abbreviations: AJCC, American Joint Committee on Cancer.
Three patients had a calculated stage 0 (T0N0M0) disease and are not included in the breakout above.
Table 2 lists the initial cancer treatments and supportive care. The most common treatment was radiation given concurrently with either chemotherapy (CRT) (30%) or biologic agent therapy (BRT) (cetuximab, 9%) followed by postoperative radiotherapy (PORT), either with chemotherapy (14%) or without chemotherapy (7%). Other treatment strategies were radiation alone (16%), induction chemotherapy followed by CRT/BRT (13%) or radiation (3%), and systemic therapy alone (2%). The remaining 6% of patients received a variety of combinations of radiation and chemotherapy. Compared with the entire population, a higher percentage of patients receiving chemotherapy alone had distant metastases at diagnosis (32% vs 2%).
Table 2. Initial Cancer Treatment and Supportive Care Received
Note that some totals may exceed 100% because of rounding.
Biologic therapy indicates cetuximab.
This category included 158 patients who could not be categorized and 6 patients who fell into multiple categories. Most of these patients received RT with chemotherapy but did not fit criteria for the predefined categories listed.
Participated in a clinical trial for head and neck cancer
N = 3598
n = 2296
n = 1302
Compared with the whole population, relatively more patients with oral cavity tumors received PORT (39% vs 21%), and less received CRT (16% vs 30%) or induction chemotherapy followed by CRT/BRT (7% vs 13%). A smaller proportion of patients with oropharyngeal cancer received radiation alone (9%) or adjuvant radiotherapy (4%), and a greater proportion received CRT (36%). Almost 33% of patients with laryngeal cancer received radiation alone, and another 33% received CRT or BRT.
Table 2 also indicates that intensity-modulated radiation therapy (IMRT) has largely replaced 3-dimensional (3-D) conformal or conventional radiation techniques, particularly in academic centers (88% vs 78%; P < .0001). The median radiation dose, number of fractions, and duration were 70 grays (Gy) (interquartile range, 66-70 Gy), 33 fractions (interquartile range, 30-35 fractions), and 7 weeks, respectively. Once-daily radiation (5 fractions per week) was the most common fractionation schedule used. Concomitant boost was used more frequently in an academic setting (15% vs 10% of patients; P < .0001), and its use did not change significantly from 2006 to 2010. Twice-daily radiation (10 fractions per week) was given infrequently (3% of patients) and did not differ between settings.
Single-agent cisplatin was most common drug (49%) and was prescribed more often in academic centers than in the community (53% vs 43% of patients; P < .0001). The drug was given most often every 3 to 4 weeks (71% of patients) and weekly (14%). Cetuximab was the next most common drug (19% of patients); and, in contrast to cisplatin, was prescribed more frequently in the community (24% vs 17% of patients; P = .0001). Multiagent chemotherapy regimens were prescribed less frequently.
More than half of patients (58%) received a feeding tube, either before or during treatment (Table 2). A feeding tube tended to be placed more often in patients who were treated at academic centers and particularly during treatment. The majority of patients were prescribed opioid analgesics and antiemetics during treatment (80% and 78% of patients, respectively) (Table 2). The use of antiemetics in patients who were receiving chemotherapy was greater (92%) than that in the overall population. Prescribing of opioids and antiemetics was more frequent in patients who were treated in academic settings.
Differences in Patterns of Care
Multivariable analyses were undertaken to test the first hypothesis that, “There is no difference in the pattern of care between community and academic settings for HNC patients.” Several differences were observed (Table 3). Characteristics associated with the likelihood of being treated in the community included age ≥70 years (there was a general pattern of bias toward an academic setting until approximately age 70 years); larynx and oral cavity cancers; treatment with BRT, CRT, or induction chemotherapy followed by radiation; placement of a feeding tube at any time; and prescribing of amifostine. Factors associated with patients being more likely to be treated at academic sites included non-Caucasian race, stage III/IV disease at diagnosis, a Zubrod performance status of 1 or 2, treatment with radiation after initial surgery, placement of a tracheotomy tube at any time, and prescribing of opioid analgesics and antiemetics.
Table 3. Logistic Regression Model of Factors associated with Treatment in Community Setting, N = 2953
Abbreviations: AJCC, American Joint Committee on Cancer; BRT, biologic radiotherapy; CI, confidence interval; CRT, chemoradiation; OR, odds ratio; RT, radiotherapy.
The reference categories were: stage, I/II; primary tumor site, other; Zubrod performance, 0; and overall treatment strategy, other (includes not assigned and RT→chemotherapy).
The intercept for the model is 0.9463.
Age: ≥70 y
AJCC stage at diagnosis: Stage III/IV
Primary tumor site
Zubrod performance status
Overall treatment strategy
Feeding tube placement
Placed any time
Tracheotomy tube placement
Placed any time
Opioid analgesic, antiemetic use
Opioid analgesic prescribed
Other supportive medications
Impact of Major Clinical Trials
To test the second hypothesis that, “The results of major randomized clinical trials will change the pattern of care in community and academic settings within 1 year of publication in peer-reviewed journals,” use of cetuximab and TPF was examined (Table 4). There was no significant change in the use of the agents/regimens after the publication of trial results in high-impact journals.
Table 4. Use of Agents Before and After Publication of Major Phase 3 Trials in High-Impact Journals
No. of Patients (%)
Abbreviation: ±, with or without; NA, not applicable; PF, cisplatin plus 5-fluorouracil; RT, radiotherapy; TPF, taxane, cisplatin, and 5-fluorouracil.
The first patient was enrolled in the Longitudinal Oncology Registry of Head and Neck Carcinoma; in December 2005; therefore, 2006 was set as the baseline.
Other induction chemotherapy regimens included single-agent cetuximab, single-agent cisplatin, carboplatin/paclitaxel, cisplatin/docetaxel, cisplatin/docetaxel/5-FU, and a variety of others.
The chi-square statistic and associated P value could not be calculated because no patients received PF. For the comparison of induction chemotherapy, other, and TPF, P = .49.
Bonner 200610 (published February 9, 2006): RT ± cetuximab
The median follow-up was 1.1 years for patients who were included in the survival-based analyses, and the maximum follow-up was 4.3 years. Locoregional control and overall survival are illustrated in Figure 3. The 1-year, 2-year, and 3-year rates of locoregional control were 95%, 93%, and 92%, respectively; and the 1-year, 2-year, and 3-year rates of survival were 86%, 76%, and 69%, respectively. The median survival was 48 months.
LORHAN is the largest repository of detailed treatment data on patients with HNC in the United States. The registry has provided important information on patterns of care in everyday clinical practice, both for primary treatment of disease and for supportive care, and it has revealed differences in patient demographic and treatment patterns between community and academic practice settings. Patients enrolled in LORHAN who were treated at academic centers tended to be younger, had more advanced disease, had fewer laryngeal cancers, and had a poorer performance status compared with patients who were treated in the community.
Overall, CRT, given either upfront or after surgery, was prescribed most for patients with newly diagnosed HNC. The combination of IMRT plus single-agent cisplatin was received frequently. Few patients received multidrug regimens, either as induction or with radiation. Most patients received a feeding tube, opioid analgesics, and antiemetics. Patterns of care at academic settings included the receipt of adjuvant CRT/BRT, IMRT, single-agent cisplatin, TPF/cetuximab, tracheotomy, opioid analgesics, and antiemetics. Patients who were treated in the community more often received BRT, radiation alone, conventional and 3-dimensional conformal radiation, single-agent cetuximab, TPF, PF, and amifostine. These differences may be explained in part by differences in baseline patient characteristics, although several of these differences persisted after controlling for important covariates. Differences in receipt of IMRT may be explained by a differing level of access to new radiation technologies in the community because of the substantial training and upfront investment required. The higher use of cetuximab and amifostine in the community suggests that similar barriers do not exist for the adoption of new drugs. It is possible that physicians in academic settings viewed the data for these drugs differently than those in the community. The higher use of tracheotomy in academic settings may reflect the preferential referral of patients with airway compromise to these centers, although it is interesting to note that more patients with laryngeal cancers were treated in the community, and the proportion with hypopharyngeal tumors was equal between the 2 settings.
Broad adoption of treatments from major trials within 1 year of publication in high-impact journals was not observed. In the cetuximab trial,10 analyses may have been hampered by the lack of a true baseline period and an insufficient sample size. It is possible that acceptance of cetuximab was heavily influenced after the presentation of the data at the American Society for Clinical Oncology meeting in June 2004,14 18 months before the first patient was enrolled in LORHAN. By the time the publication came out in February 2006, the study results may have been perceived as “old news.” The distribution suggests that cetuximab probably is prescribed when contraindications for cisplatin exist.
It is rather surprising that the 3 major positive trials on TPF did not appear to broadly influence daily clinical practice within 1 year of their publication. A plausible explanation is that these scientifically important trials produced a treatment option that can be adopted as an alternative to, rather than as a replacement for, the already well established radiation-cisplatin platform in the United States. It is therefore likely that most oncologists are awaiting the results of phase 3 trials comparing induction TPF followed by CRT versus concurrent radiation-cisplatin alone before routinely adopting induction TPF into their practice. It is also possible that the additional toxicity of a taxane, although manageable in the context of a clinical trial, was more problematic in busy daily clinical practice.
The generalizability of the findings from LORHAN depends on the representativeness and quality of the data. Enrollment in LORHAN required written informed consent by the patient; and, although participating physicians were encouraged to approach all eligible patients, we do not know how many patients declined or whether and how these patients differed from those who participated. Another potential bias is the positive selection of physicians with a self-declared interest in HNC or physicians with a bias in favor of cetuximab, a product marketed by the registry sponsor, Eli Lilly and Company. However, the distribution and characteristics of cases suggest that patients participating in LORHAN are similar to those in the Surveillance, Epidemiology, and End Results Program. A strength of LORHAN is that few patients were lost to follow-up or withdrew consent. The most common reason for registry discontinuation was death. Data logic and consistency were monitored on an ongoing basis by the LORHAN advisory board. Reports of outliers were generated and sites were contacted to confirm data points, as needed. Although information entered into LORHAN was not audited against source documents, these reviews provide some measure against systematic error. Random error is likely addressed by the size of LORHAN. Interpreting the outcomes of LORHAN would have benefited from information on the quality of radiation planning and delivery, but such efforts require substantial funding to complete.
In summary, LORHAN has documented differences in the treatment of patients in community and academic settings for a series of patients in the United States. The establishment of similar registries outside the United States would provide the means for geographic comparisons and should be pursued. Future analyses of LORHAN will focus on questions that can be addressed uniquely by the registry, such as disparities in care and shifting treatment paradigms in specific diseases like oropharyngeal cancer.
LORHAN is funded by Eli Lilly and Company (formerly by ImClone Systems Inc.).
CONFLICT OF INTEREST DISCLOSURES
K. Kian Ang is compensated as a member of the Eli Lilly Advisory Board. Amy Chen and Lisa A. Bellm act as consultants in an advisory role for MedNet Solutions, Inc. Marc Schwartz is an employee of MedNet Solutions. Douglas Adkins has received research funding and honoraria from Eli Lilly.