Kaplan-Meier overall survival curves by radiotherapy (RT) dose.
Kaplan-Meier overall survival curves by chemotherapy.
Pedruzzi PAG, Kowalski LP, Nishimoto IN, Oliveira BV, Tironi F, Ramos GHA. Analysis of Prognostic Factors in Patients With Oropharyngeal Squamous Cell Carcinoma Treated With Radiotherapy Alone or in Combination With Systemic Chemotherapy. Arch Otolaryngol Head Neck Surg. 2008;134(11):1196–1204. doi:10.1001/archotol.134.11.1196
To assess the prognostic significance of several factors in oropharyngeal squamous cell carcinoma treated with radiotherapy alone or in combination with chemotherapy.
Erasto Gaertner Hospital, Curitiba, Brazil, and A. C. Camargo Hospital, São Paulo, Brazil.
A total of 361 patients treated for squamous cell carcinoma from January 1, 1990, to December 31, 2001.
Radiotherapy alone or with chemotherapy.
Main Outcome Measures
Disease-free survival, overall survival, and treatment response.
Most tumors were located at the tonsil (46.8%) or base of the tongue (28.0%) and were at clinical stage III or IV (92.8%). Treatment response was associated with Zubrod scale score, weight loss, number of comorbidities, symptom-severity and Piccirillo stages, hemoglobin level, tumor site, macroscopic appearance of the tumor, and clinical stage. The 5-year overall survival rate was 17.6% and disease-free survival rate was 16.2%. The significant prognostic variables were age; Zubrod scale score; weight loss; comorbidities; Berg, Piccirillo, and symptom-severity staging; involvement of adjacent soft-tissue areas and bone; lymph node mobility; clinical stage; and radiotherapy doses. The multivariate analysis showed Zubrod scale score, symptom-severity staging system, Berg staging system, comorbidities, and radiotherapy dose as independent prognostic factors.
A combination of clinical factors, such as symptoms, patients' general status, weight loss, and comorbidities, leads to a relevant stage of clinical severity that can be associated with the TNM stage as predictors of survival in oropharyngeal carcinoma.
Malignant neoplasms of the oral cavity and pharynx are among the major public health problems worldwide, with more than 400 000 new cases diagnosed every year.1,2 According to American Cancer Society statistics, 34 360 new cases and 7550 deaths were expected in the United States in 2007.2 On the basis of a Surveillance, Epidemiology, and End Results database study,1 no changes were observed for the incidence rates of oropharyngeal carcinomas in the US population from 1974 to 1999. Around 70% of patients were diagnosed as having advanced clinical stage (regional and distant stages in the Surveillance, Epidemiology, and End Results classification). Although radiotherapy alone was the most frequently adopted treatment, there was a significant decrease in its use from 54.2% to 38.0% during the period from 1974 through 1997. In the same period, surgery alone was adopted in about 18% of the cases and combined surgery and radiotherapy was used with increasing frequencies from 21.0% to 34.2%. Following the treatment changes there was a significant increase in 5-year survival probabilities from 36.3% for 1974 through 1976 to 49.1% for 1995 through 1997.1
Most studies on prognostic factors in oropharyngeal cancer include other pharyngeal or oral cavity tumors. Tumor extension has traditionally been assessed through TNM staging, and it has been used as the main variable for dividing patients into prognostic groups.3 However, several authors have proposed other options of grouping the TNM staging system to improve its prognostic significance.4- 9 Besides tumor extension, other clinical and pathological variables, not included in the TNM system, have been identified as prognostic predictor variables. More recently, clinical factors such as the patient's general status, symptoms, and comorbidities have been described as significant prognostic factors in functional staging.10- 13
Identifying predictors of treatment response and prognostic factors is important for the evaluation of cancer recurrence risk and treatment planning.3 Because large prospective studies comparing surgery and radiotherapy have not been conducted, the retrospective analysis of prognostic factors has been applied for treatment selection in new patients with oropharyngeal cancer.14 The objective of this retrospective study was to assess, in patients with oropharyngeal squamous cell carcinoma subjected to radiotherapy alone or in combination with chemotherapy, factors associated with treatment response and prognosis.
This study is based on a retrospective analysis of all patients with primary squamous cell carcinoma of the oropharynx who underwent radiotherapy alone or in combination with chemotherapy from January 1, 1990, to December 31, 2001, at Erasto Gaertner Hospital and A. C. Camargo Hospital in Brazil. All patients' information was collected from the medical records by the same investigator (P.A.G.P.).
The inclusion criteria were oropharyngeal squamous cell carcinoma not previously treated and not presenting with distant metastases at diagnosis. Demographic, clinical, and treatment variables were studied. Weight loss was considered present when it represented at least 10% of the previous weight in a period of up to 6 months.
Symptoms were grouped according to their severity. Initially we adapted the model devised by Piccirillo et al15 (referred to as the Piccirillo stage) for grading severity of laryngeal cancer symptoms (Table 1). We also used a model based on that proposed by Pugliano et al,11 categorizing symptoms as local, extralocal, and regional (Table 1). This model, referred to as the symptom-severity stage, included 4 symptoms (weight loss, dysphagia, cervical mass, and otalgia), and 4 severity stages (none [absence of all 4 symptoms], mild [1 of the 4 symptoms], moderate [2 of the 4 symptoms], and severe [3 or 4 of the 4 symptoms]).
For hemoglobin level, we used the value measured immediately before radiotherapy. The cutoff used was at 12.5 g/dL in accordance with the findings of Denis et al.16 (To convert hemoglobin to grams per liter, multiply by 1.) The clinical staging system adopted was the fifth edition of the TNM manual.17 The other assessed staging methods were those of Berg,4 Jones et al5 (T and N integer score), Snyderman and Wagner,6 Hart et al,7 Kiricuta,8 and Hall et al.9
The patients who had not finished the planned treatment were kept in the study for the intention-to-treat analysis. Treatment response at the end of the treatment was divided into the following categories: no response, partial response, and complete response. The first category included patients who experienced a progression in the tumor stage or showed no objective response. Partial response was assigned when tumors or lymph nodes diminished by at least 50% on clinical or imaging examinations. Finally, the response was considered complete when patients presented no documented evidence of the disease once the treatment was concluded and remained free of disease for at least 3 months.
The statistical analysis was carried out by means of the statistical package Stata (version 7.0; StataCorp, College Station, Texas). To verify the relationship between the treatment response and the study variables, the χ2 test was used when all expected frequencies in contingency tables were greater than 5. The survival probabilities were estimated by the Kaplan-Meier method and the comparison of survival curves by the log-rank test. The follow-up period was the time, in months, between the date of the start of treatment and the date of last follow-up information or the patient's death. Similarly, to calculate the disease-free survival rate, we considered the interval between the beginning of treatment and the date of the diagnosis of the first recurrence. Because some patients had no response to treatment, the beginning date was the choice for this analysis. The 5% level of significance was used for all statistical tests. All variables with log rank P < .20 in the univariate analysis were candidates for Cox regression multivariate analysis to estimate the adjusted hazard ratio and 95% confidence interval. The independent prognostic factors were based on the estimated hazard ratio and 95% confidence interval from the multivariate Cox regression model. The treatment response was not included in the multivariate analysis because the analysis included known prognostic factors.
A total of 361 cases of oropharyngeal squamous cell carcinoma were evaluated. Of these, 322 patients were male (89.2%) and 39 female (Table 2). The patients' ages ranged from 27 to 85 years (mean, 57 years). On the Zubrod scale, 196 cases (54.3%) were classified as 0/1, 93 (25.8%) as 2, and 18 (5.0%) as 3. Weight loss was observed in 243 patients (67.3%). The hemoglobin levels before radiotherapy ranged from 5.1 to 17.8 g/dL (mean, 13.2 g/dL). Of the patients with information, most were tobacco smokers (94.9%) and alcohol drinkers (79.8%). Comorbidities were not observed in 152 patients (42.1%). Pulmonary diseases were the most frequent comorbidities (154 cases [42.7%]), followed by cardiac diseases in 76 (21.1%). Odynophagia was the most common symptom (211 cases [58.4%]) followed by enlarged neck nodes (149 cases [41.3%]).
Radiotherapy alone was used in 264 patients (73.1%) and radiotherapy combined with chemotherapy in 97 patients (26.9%). Chemotherapy was neoadjuvant in 62 patients (17.2%) and concomitant in 35 patients (9.7%). Cisplatin combined with fluorouracil was used in 60 patients, cisplatin alone in 22 patients, and other drug combinations in 15 patients. The dose of radiotherapy ranged from 12 to 79 Gy (median, 64.19 Gy). For 68 patients (18.8%), the dose received was less than 60 Gy; for 67 patients (18.6%), it was between 60 and 69 Gy; and for 226 patients (62.6%), it was 70 Gy or greater. A total of 313 patients (86.7%) completed the planned treatment, whereas 48 patients (13.3%) did not. Radiotherapy, alone or combined with chemotherapy, was delivered by a 4- to 6-MeV linear accelerator.
A total of 102 patients (28.3%) achieved a partial response, 132 patients (36.6%) had a complete response, and 127 patients (35.2%) either did not show any response or experienced disease progression during the treatment. The response rates were significantly influenced by Zubrod scale score, weight loss, number of comorbidities, Piccirillo stage, symptom-severity stage, and site and macroscopic appearance of the primary tumor (Table 2). The association between the treatment response and the staging system is shown in Table 3.
We observed a higher rate of complete response in patients whose radiotherapy dose was greater than 70 Gy (<60 Gy, 6 of 68 patients [8.8%]; 60-69 Gy, 27 of 67 [40.3%]; and ≥70 Gy, 99 of 226 [43.8%]) (P < .001). Concomitant or neoadjuvant chemotherapy type was not statistically significant in the analysis of treatment response, showing the following response rates (partial plus complete): no chemotherapy, 64.8%; neoadjuvant, 62.9%; and concomitant, 68.5%.
After a follow-up time of 2 days to 122 months (mean, 24.7 months), the last objective information available on each patient showed that 289 patients (80.1%) had died of persistent or recurrent disease, 36 (10.0%) were alive and disease free, 14 (3.9%) had died of other causes, and 22 (6.1%) had no updated follow-up information. The mean time to recurrence was 18 months. The 5-year overall survival (OS) rate was 17.6%, and the 5-year disease-free survival (DFS) rate was 16.2%.
Patients 45 years or younger as well as those older than 65 years had the lowest OS rates, but this variable was not significantly associated with DFS. Statistically significant differences in OS and DFS rates were associated with Zubrod scale score. None of the patients with a Zubrod scale score of 3 survived for 5 years. Patients with weight loss, comorbidities, nonlocal classification on the Piccirillo scale, symptom-severity scale stage scored as anything other than none, and tumor extension to soft tissues or bone had the lowest survival rates. Also, the OS was higher in the patients with mobile lymph nodes than in those with semifixed or fixed neck metastasis (Table 4). All of the stage classifications were significantly associated with the survival rates (Table 5).
The survival rates were significantly higher for the patients who concluded the planned radiotherapy (complete radiotherapy, 20.3% OS at 5 years; incomplete radiotherapy, 0%), as well as for those whose radiotherapy doses were 70 Gy or greater (<60 Gy:, 2.9% 5-year OS; and 60-69 Gy, 14.2%; ≥70 Gy, 22.4%) (P < .001) (Figure 1). Chemotherapy did not have a statistically significant effect on OS rates (no chemotherapy used, 17.4% 5-year OS; neoadjuvant chemotherapy, 17.3%; and concomitant chemotherapy, 19.2%) (Figure 2).
The crude and age-adjusted relative risks of death and their respective 95% confidence intervals obtained by Cox regression model are shown in Table 6. Estimated independent prognostic factors in this study were Zubrod scale score, symptom-severity stage, presence of comorbidities, Berg stage, and radiotherapy dose.
Oropharyngeal cancer is usually diagnosed at advanced clinical stages (approximately 70% of the reported cases).1,14,18,19 Indeed, in the present study, 92.8% of the cases were diagnosed at advanced stages. The 5-year OS probability was 17.6% and was slightly higher (20.3%) for the patients who completed the planned treatment. This survival rate is rather low compared with the 5-year survival rates for patients with stages III and IV disease subjected to conventional surgical or radiotherapy treatments (30%-50%).1,18,20 Hart et al,7 analyzing data from 640 patients with oropharyngeal squamous cell carcinoma, reported a survival rate of 28%. Comparing both studies, however, it is important to emphasize that in our series we excluded patients who had undergone surgery, whereas the study carried out by Hart et al7 included patients who underwent different curative treatments. Another major factor is that in our study the patients who did not complete the planned treatment remained in the survival analysis to obtain the actual results of this group of patients.
The male patients had lower survival rates than did the female patients; indeed, a number of studies have shown that women have a better prognosis.7,19,21 In the present study, we did not observe significantly different OS rates between women (25.5%) and men (16.6%).
The functional effects of the tumor on the patient may be assessed by means of symptoms, the patient's general status, and comorbidities.10- 13,15 At diagnosis, patients with several types of cancer can present with functional alterations that impair their physical capabilities or general medical conditions.10,13 A number of studies have shown the importance of the clinical status of the patient in the assessment of survival rates and radiotherapy response.13,22 Weight loss is a significant predictor of mortality in several neoplasias, including oral and oropharyngeal cancers. The loss of more than 10% of the patient's weight before treatment is a significant prognostic factor.22 In the present study, both the Zubrod scale score and weight loss were significant predictors of treatment response, OS, and DFS.
Morphologic description of the disease does not always provide information on the severity of the tumor.10 By adapting Piccirillo and coworkers' model,15 applied in this study to oropharyngeal cancer, we observed an association between symptom severity and OS.
Pugliano et al11 emphasized the same findings as Piccirillo et al,15 that localized tumors do not always present with only localized symptoms. In this study, Pugliano and coworkers' model was shown to be a significant predictor of treatment response and survival. In the multivariate analysis, this model was demonstrated to be an independent prognostic variable. Ribeiro et al,13 when studying oral cancer, adapted Pugliano and coworkers' model and also found an association between this model and prognosis. The results of the current study as well as the one by Piccirillo et al15 pointed out that it is necessary to improve the current staging systems by using symptoms as variables to better predict survival in treated patients. The prognosis of patients with oropharyngeal cancer may be determined accurately on the first medical visit by assessing the symptom severity in addition to the presence of weight loss. The presence of varied and significant symptoms in patients with early-stage tumors may point toward greater tumor aggressiveness.
Comorbidities are defined as concomitant diseases not related to the disease under investigation.10- 12 In the present study, 53.2% of the patients presented with 1 or 2 comorbidities, and this figure is comparable to those reported by Piccirillo,12 who found comorbidities in 45% of the patients with head and neck cancer. Comorbidities are important because they can have a tremendous effect on the patient's care, including initial treatment selection, treatment tolerance, and adverse effects. Piccirillo12 described the association of comorbidities with prognosis in patients, which was confirmed by other studies.11,13 In the present study, we also observed a significant association between the presence of comorbidities with treatment response, OS, and DFS.
Lower tissue oxygenation is usually considered to be a factor that diminishes the response to radiotherapy. The hemoglobin level before the start of treatment is a prognostic marker of the response to chemotherapy and radiotherapy in patients at advanced stages of head and neck cancer.21 A study carried out by the French Head and Neck Oncology and Radiotherapy Group showed that, of the 226 patients with oropharyngeal squamous cell carcinoma treated with radiotherapy alone or combined with chemotherapy, those who had hemoglobin levels less than 12.5 g/dL before treatment had a lower rate of locoregional control of the disease, as well as of survival.16 In the present study, we observed only a tendency for correlation between the hemoglobin levels and OS rates, possibly because most patients with a hemoglobin level less than 10 g/dL received blood transfusions.
The TNM stage is a major prognostic factor in most studies on head and neck cancer.10- 23 In the present study, the T and N clinical stages correlated with OS and DFS, similar to the findings reported in other studies.10,11,20,21,23 All of the proposed systems of grouping TNM categories4- 9 were significantly associated with survival rates in univariate survival analysis. The discrimination ability of each staging system was then evaluated by means of the multivariate Cox model. The Berg4 system was identified as the best predictor of survival among all of the staging systems. This system is the same one proposed by the TNM classification.
The current study showed that lymph node mobility at physical examination was an important prognostic factor for 5-year OS. Patients with fixed lymph nodes had one of the lowest long-term survival rates observed in the population studied. Unfortunately, these data cannot be compared with the literature because mobility of lymph nodes was dropped from the International Union Against Cancer and American Joint Committee on Cancer staging systems since the 1970s because of disagreement about its clinical evaluation. Possibly, variables such as invasion of carotid, skull base, or prevertebral tissues that causes fixation should be considered for description of fixation in agreement with the International Union Against Cancer and American Joint Committee on Cancer systems because it is much less susceptible to misinterpretation.
The selection of patients for radiotherapy has been a challenge because there are no reliable criteria for the prediction of treatment response. Tumor site, macroscopic appearance, and tumor extension have been described as possible predictors of tumor response to treatment.23 In the present study, we showed a better response for soft palate (71.4%) and tonsillar (63.9%) tumors than for tumors of the base of the tongue (62.4%). In addition, exophytic tumors had a better response (75.0%) than endophytic tumors (60.2%). Similarly, the survival rates were higher for soft palate and uvula cancers (27.1%), followed by tumors of the tonsil (1.1%) and base of the tongue (15.0%).
The results of this study are disappointing when compared with the most recent literature on chemoradiotherapy. These differences deserve future investigation. Possibly we are dealing with different carcinogenesis of oropharyngeal squamous cell carcinomas. In several parts of the world, human papillomavirus is a major risk factor. However, in our population the prevalence of human papillomavirus infection is very low.24 The main risk factor in our population is heavy smoking and drinking (specially hard liquor and sugar cane spirits [cachaça]).25 These patients usually have significant comorbidities and weight loss that can explain the poor results.
There are some controversies about the use of combined radiotherapy and chemotherapy or surgery and postoperative radiotherapy in treating advanced oropharyngeal cancer.1,14,16 A number of level I evidence studies have shown that the association of radiotherapy and chemotherapy can improve the survival results when compared with radiotherapy alone.16,20,26 Most of them include only patients who have high performance status, which is not the common situation for patients with advanced disease diagnosed in socioeconomically disadvantaged populations. In the present study, we did not observe any difference in survival rates according to the treatment (radiotherapy alone or in combination with neoadjuvant or concomitant chemotherapy). Patients with advanced disease and with performance status impairments like those included in this study should probably be considered in future clinical trials aiming to introduce advances in the treatment of socially disadvantage patients living in geographic areas outside the developed world.
Correspondence: Luiz P. Kowalski, MD, PhD, Department of Head and Neck Surgery and Otorhinolaryngology, A. C. Camargo Hospital, Rua Professor Antonio Prudente 211, 01509-900 São Paulo, Brazil (firstname.lastname@example.org).
Submitted for Publication: April 25, 2007; final revision received January 29, 2008; accepted February 4, 2008.
Author Contributions: Dr Pedruzzi had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Pedruzzi, Kowalski, and Ramos. Acquisition of data: Pedruzzi, Oliveira, and Tironi. Analysis and interpretation of data: Pedruzzi, Kowalski, and Nishimoto. Drafting of the manuscript: Pedruzzi. Critical revision of the manuscript for important intellectual content: Pedruzzi, Kowalski, Nishimoto, Oliveira, Tironi, and Ramos. Statistical analysis: Nishimoto. Obtained funding: Pedruzzi. Administrative, technical, and material support: Pedruzzi, Oliveira, Tironi, and Ramos. Study supervision: Pedruzzi and Kowalski.
Financial Disclosure: None reported.
Previous Presentation: This study was presented at the annual meeting of the American Head and Neck Society; April 30, 2007; San Diego, California.