Algorithm for determination of the extended clinical severity staging system. ASA indicates American Society of Anesthesiologists; BMI, body mass index (calculated as weight in kilograms divided by the square of height in meters); NCI, National Cancer Institute; and FSI, Functional Severity Index.
Five-year overall survival curves according to the extended clinical severity staging system.
Five-year overall survival curves according to the type of complications.
Ribeiro KDCB, Kowalski LP, Latorre MDRDDO. Perioperative Complications, Comorbidities, and Survival in Oral or Oropharyngeal Cancer. Arch Otolaryngol Head Neck Surg. 2003;129(2):219-228. doi:10.1001/archotol.129.2.219
To establish the value of clinical factors in the prediction of perioperative complications and death in patients with oral and oropharyngeal carcinomas and to develop a new extended clinical severity staging system that combines patient and tumor factors.
Patients and Methods
A total of 530 patients with oral or oropharyngeal carcinomas submitted to surgical treatment were studied. Logistic regression was used to identify risk factors for perioperative complications, and the Cox proportional hazards regression model was used to establish independent prognostic factors.
Daily alcohol consumption, smoking, sex, neck lump, earache, pain, dysphagia, weight loss, oral bleeding, odynophagia, body mass index, National Cancer Institute comorbidity index score, American Society of Anesthesiologists surgical risk, hematocrit level, and total lymphocyte count had an impact on prognosis in univariate analysis. Survival according to extended clinical severity stage was 76.7% for stage 1, 64.4% for stage 2, 44.8% for stage 3, and 25.5% for stage 4 (χ2 = 64.16; P<.001). In multivariate analysis, only APACHE II score, neck dissection, POSSUM index score, and type of reconstruction were independent risk factors for perioperative complications. The final prognostic model included development of local plus systemic complications, extended clinical severity stage, type of reconstruction, and APACHE II score.
Clinical variables have a predictive effect on morbidity and mortality of patients with oral and oropharyngeal cancer treated surgically. Local plus systemic perioperative complications can adversely affect the prognosis. The uniformity of results confirms that survival estimates can be enhanced by the addition of clinical characteristics to the TNM classification, creating a more accurate system for the estimation of prognosis.
ORAL AND oropharyngeal cancers are predominant in the older population. Surgery and radiotherapy are the primary therapeutic procedures, and the choice of therapy depends on factors related to the tumor, the patient, and institutional experience.1 Anatomic extension of the disease, described through the TNM staging system,2 has long been accepted as one of the most important prognostic factors. However, recent studies suggest that symptoms, comorbidities, and other clinical characteristics of the patients are important for therapeutic planning and for determining the risk of complications3 and the prognosis of several types of cancer.4- 10 The addition of these factors to the TNM classification permitted the creation of new staging systems, superior in the prediction of survival compared with the TNM staging system alone.5- 7,11- 13
Recent advances in anesthetic and surgical techniques, including microvascular reconstruction, have allowed the use of more radical oncologic procedures for advanced head and neck cancer. However, the curative intent can be limited by the hazard of life-threatening perioperative complications. The occurrence of perioperative complications increases the length of hospital stay and the need for other diagnostic and therapeutic procedures, with a subsequent increase in hospital costs.14 Local plus systemic complications and an advanced clinical severity stage of disease are responsible for a worse prognosis in patients with oral cancer.3 There are no preoperative scoring systems directed at particular types of surgery, and reported results are based on studies including all types of surgical procedures.14 Accurate staging and prediction of perioperative complications are of paramount importance for treatment planning in elderly or chronically ill patients with a variety of pathophysiologic alterations because they require continuous medical assistance and high technologic support. However, assessment of the results of providing such care is limited and imprecise. Furthermore, a severity of disease classification system is essential to estimate the pretreatment risk of death, the appropriate indication of surgical treatments, and the prediction of outcome.15
The objectives of this study are to evaluate the importance of clinical factors in the prediction of perioperative complications and survival in patients with oral cavity and oropharyngeal cancer and to develop a more accurate estimate of the prognosis by combining patient and tumor factors in a new staging system.
The medical records of 530 patients with squamous cell carcinoma of the oral cavity and oropharynx admitted to the Centro de Tratamento e Pesquisa Hospital do Câncer A. C. Camargo between January 1, 1990, and December 31, 1997, were reviewed. The following criteria were used for inclusion in the study: a histologically confirmed diagnosis, absence of previous oncologic treatment for this primary tumor, no distant metastasis, and surgical treatment with a curative purpose, exclusive or as part of a multidisciplinary approach.
Data collection from the medical records was performed using a specially designed form. These data included demographic information, hematocrit level, hemoglobin level, body mass index, smoking status, alcoholism status, TNM staging (Union Internationale Contre le Cancer or American Joint Committee on Cancer classification),2 tumor site, surgical risk according to the American Society of Anesthesiologists (ASA),16 type of surgery and neck dissection, type of reconstruction, blood transfusion, comorbidities according to National Cancer Institute classification,17 Charlson Comorbidity Index score,18 Functional Severity Index,13 extended clinical severity stage (ECSS),13 APACHE II (Acute Physiology and Chronic Health Evaluation II) score,19 POSSUM (Physiological and Operative Severity Score for Enumeration of Mortality and Morbidity) index score,20 length of stay in an intensive care unit (ICU), length of hospital stay, and perioperative complications. Outcome measures included development of complications in the immediate postoperative period (30 days) and 5-year overall survival rates. Patients were followed from the date of diagnosis to the date of last objective evaluation or death. Only 9.1% of the patients were lost to follow-up.
The APACHE II classification is a revised version of a prototype system, the APACHE, and includes 12 physiologic measures (temperature; mean arterial pressure; heart rate; respiratory rate; oxygenation; arterial pH; serum sodium, potassium, and creatinine levels; hematocrit level; white blood cell count; and Glasgow Coma Scale score), age, and severe chronic health problems. The physiologic score is determined from the worst value, for example, the lowest hematocrit level or the highest respiratory rate, during the initial 24 hours after ICU admission (Table 1).19 We also applied the APACHE II for patients not referred to the ICU based on findings from the first 24 hours after surgery.
The POSSUM index was developed by multivariant discriminant analysis to obtain a method of risk assessment. A 12-factor, 4-grade physiologic score was developed that included age; cardiac status; pulse rate; systolic blood pressure; respiratory status; Glasgow Coma Scale score; serum concentrations of urea, potassium, and sodium; hemoglobin concentration; white blood cell count; and findings on electrocardiography. This score was combined with a 6-factor operative score that compensates for the type of surgical procedure and includes type and number of procedures, volume of blood loss, peritoneal contamination, presence and extent of malignancy, and timing of surgery (Table 2).20 In this study, operative severity was minor (pelvectomy, partial glossectomy, or intraoral resection without neck dissection), moderate (total glossectomy, hemiglossectomy, pelviglossectomy, sectional pelviglossomandibulectomy, marginal pelviglossomandibulectomy, wide excision, classic or modified retromolar operation, inframesostructure resection, buccopharyngectomy, or any surgical procedure with unilateral neck dissection), major (any surgical procedure with myocutaneous flap reconstruction), or major plus (any surgical procedure with microvascular reconstruction). For both indexes, if an item was not evaluated and the patient did not have any disease that could be responsible for an abnormal result, the lowest score was assigned.
Patients were electively referred to the ICU based on one of the following criteria: expected operative time longer than 6 hours or the presence of a comorbidity diagnosed before surgery, such as coronary insufficiency or advanced chronic obstructive pulmonary disease requiring strict cardiopulmonary monitoring.
Wound infection was recorded only for patients with suppurative drainage and those who developed a mucocutaneous fistula.21 Wounds noted to have erythema or erythema and edema may be considered to be infected in the presence of fever, based on an evaluation performed by the surgeon.
The information from the forms was entered into a database (Dbase for Windows; Borland International, Scotts Valley, Calif). Periodically, revisions were made to verify the internal consistency of the data. For the statistical analysis, commercially available software (SPSS for Windows, release 10.0; SPSS Inc, Chicago, Ill) was used. Descriptive statistics were used as a preliminary analysis of the relation between baseline variables and outcome events. The t test was used to compare means. Continuous variables were categorized to facilitate data analysis and presentation. Logistic regression was used to find independent risk factors for perioperative complications (present or absent). Survival analysis was performed using the Kaplan-Meier method (the log-rank test was used to compare the curves) and the Cox proportional hazards model to estimate independent risk factors for death. For all statistical tests, α = .05 was established.
The cohort of 530 patients included 439 men (82.8%) and 91 women (17.2%); 446 patients were white (84.2%) and 84 (15.8%) belonged to other ethnic groups. The mean patient age was 57.3 years (range, 27-87 years). Tumors were located in the following sites: 34 on the base of the tongue (6.4%), 138 on the oral tongue (26.0%), 52 on the gums (9.8%), 129 on the floor of the mouth (24.3%), 14 on the palate (2.7%), 71 on other parts of the mouth (13.5%), 84 on the tonsils (15.8%), and 8 on the oropharynx (1.5%). Most patients had advanced tumors (TNM clinical stage IV, 47.3%) and reported a variety of symptoms (Table 3).
All patients underwent surgery as the primary treatment, and 330 patients were exposed to radiation as adjuvant therapy. Four hundred seventy-eight patients also underwent neck dissection as a part of the initial treatment. Operative time ranged from 25 to 960 minutes (median, 360 minutes). Most patients (54.3%) did not undergo intraoperative blood transfusion. Primary closure was used for 216 patients (40.8%), and several types of reconstruction were used on the remaining patients (Table 4). Two hundred forty-eight patients (46.8%) were electively referred to the ICU 1 to 20 days (median, 2 days) after surgery. The length of hospital stay ranged from 0 to 71 days (mean, 8 days). Patients with perioperative complications had a significantly longer hospital stay (mean, 11.9 days) than those without perioperative complications (mean, 7.1 days) (P<.001).
The incidence of perioperative complications was 58.9%. Wound infection (32.5%) and dehiscence (26.2%) were the most frequent events (Table 5). Postoperative mortality was 2.6%.
Ninety-six percent of patients had comorbidities when classified according to the National Cancer Institute index compared with 39.1% when using the Charlson Comorbidity Index. The ASA classification provided the following results: ASA I, 9.3%; ASA II, 62.6%; ASA III, 27.6%; and ASA IV, 0.6%. The median POSSUM index score was 27 (range, 19-48), and the APACHE II score varied from 0 to 23 (median, 7).
The Cox proportional hazards model identified, in an univariate analysis, 15 variables that affected prognosis (P≤.10): daily alcohol consumption (hazard ratio [HR], 1.5; P = .003), smoking (HR, 1.5; P = .03), male sex (HR, 1.4; P = .04), neck lump (HR, 1.5; P = .002), earache (HR, 1.5; P = .005), pain (HR, 1.5; P = .002), dysphagia (HR, 2.3; P<.001), weight loss (HR, 1.4, P = .002), oral bleeding (HR, 2.0; P<.001), odynophagia (HR, 1.4; P = .02), body mass index (calculated as weight in kilograms divided by the square of height in meters) of 22.3 or less (HR, 1.3; P = .01), National Cancer Institute comorbidity index score greater than 2 (HR, 1.4; P = .008), ASA surgical risk III/IV (HR, 1.5; P = .003), hematocrit level less than 34.5% (HR, 1.5; P = .03), and total lymphocyte count of 1.2 × 103/µL or less (HR, 1.6; P = .008). The Functional Severity Index was built through multiplication of the HRs for each patient. When the condition was not present, we gave the value of 1 for that category. The score ranged from 1 to 199.37 (median, 8.6), and the patients were grouped into 3 categories based on terciles: high (score >12.90), intermediate (score >6.14 and ≤12.90), and low (score ≤6.14) grade of functional impairment. Survival analysis confirmed a statistically significant difference in 5-year overall survival among the 3 groups of the Functional Severity Index: low grade, 63.2%; intermediate grade, 45.8%; and high grade, 22.8% (χ2 = 52.53; P<.001). Therefore, according to a previously described method,11 the next step was the conjunction of the Functional Severity Index with the TNM staging system. The categories of the conjunction of the 2 classifications were then consolidated to create an ECSS system, also composed of 4 stages (Figure 1). Five-year overall survival for this ECSS system was as follows: stage 1, 76.7%; stage 2, 64.4%; stage 3, 44.8%; and stage 4, 25.5% (χ2 = 64.16; P<.001) (Figure 2).
The comparison among the systems demonstrated that the ECSS system overcame TNM, exhibiting a higher survival gradient (51.2 vs 37.4) and a higher χ2 test value (64.16 vs 44.33). When both variables (TNM and ECSS) were included in the multivariate models for recurrence or tumor-specific survival, statistical significance was lost. Therefore, it was decided to keep ECSS in the final models, adjusted for type of reconstruction, type of complications, and APACHE II score. In this way, the ECSS system also could be recognized as better than TNM alone in the prediction of tumor-specific survival and recurrences (Table 6).
The occurrence of perioperative complications was associated with the following variables: APACHE II score (P<.001), surgical time (P<.001), TNM clinical stage (P<.001), ECSS (P<.001), neck dissection (P<.001), type of reconstruction (P<.001), intraoperative blood transfusion (P<.001), staying in an ICU (P<.001), ASA surgical risk (P = .02), POSSUM index score (P<.001), and race (P = .02) (Table 7). In multivariate analysis, only APACHE II score (OR, 2.6; P<.001), neck dissection (unilateral: OR, 4.3; P<.001; and bilateral: OR, 4.8; P<.001), POSSUM index score (OR, 1.7; P = .02), and type of reconstruction (OR, 2.1; P = .001) were identified as independent risk factors for perioperative complications (Table 8).
The final predictive prognostic model included the development of local plus systemic complications (Figure 3) (HR, 2.2), ECSS (stage 3: HR, 2.1; stage 4: HR, 3.4), type of reconstruction (myocutaneous or microvascular flap: HR, 1.6), and APACHE II score greater than 10 (HR, 1.3) (Table 9).
Patients with head and neck cancer are especially likely to have comorbidities as a result of chronic alcohol and tobacco consumption. Comorbidities can impact the diagnosis, prognosis, and treatment of patients with cancer.22
The TNM classification has been universally accepted and widely used to describe tumor characteristics and predict survival rates. However, the system fails for not taking into account the clinical biological features of the cancer, which is expressed by structural changes and its physiologic detriments in the patient.13 The gross anatomic aspects (extent of the disease), the microscopic appearance (cell type and degree of differentiation), and the biomolecular characteristics (tumor markers and ploidy) are different modes to describe tumor morphologic structure.23 Cancer symptoms (type, duration, and severity)24 and the performance status of the host25 are clinical factors that represent the seriousness of illness in a patient. Although comorbidity is not related to the cancer itself, it is an important clinical aspect for being able to affect the choice of treatment and prognosis.6,7,10,26
The purpose of using symptoms as a prognostic factor in cancer is not new, since important prognostic information is already described for different types of cancer.6,7 Neel et al27 demonstrated that when patients with nasopharyngeal cancer were grouped into 2 categories based on the number and duration of symptoms (eg, epistaxis, loss of sense of smell, and tinnitus) at initial presentation, they had significantly different survival rates in each category. In patients with osteosarcoma, weight loss had a notable prognostic impact that remained important in multivariate analysis even after controlling for other possible prognostic factors, such as symptom duration, tumor site, regional spread, grade, microscopic morphologic findings of the tumor, swelling at the tumor site, and lytic appearance of the tumor.28 The presence of chills, fevers, or night sweats has been correlated with worse survival in patients with lymphoma for many years, being used as a modifier in the staging system.29
Previous studies9,11- 13 on laryngeal, oral, and oropharyngeal cancers proved that symptom severity contributes additional prognostic data not available from anatomic staging alone. Our findings that symptoms such as a neck lump, dysphagia, weight loss, and oral cavity bleeding are prognostic factors validate results from previous studies.11- 13,29
To our knowledge, odynophagia had not yet been proved to have an independent impact on overall survival in patients with oral or oropharyngeal cancer until this publication. Pugliano et al29 demonstrated the effect of this symptom on survival rates only in a univariate analysis. However, this symptom had already been reported30 as an independent prognostic factor in patients with squamous cell carcinoma of the hypopharynx. A previous study31 reported that the presence of odynophagia was associated with advanced stages owing to lateness of diagnosis, but in our series the same frequency of the symptom has been found in the initial and advanced stages (data not shown).
Alcohol consumption remained related to survival rates, as in our previous study,13 which emphasizes that the high prevalence of alcohol abuse among these patients justifies the inclusion of alcohol use in a prognostic system.32
Smoking determined a worse prognosis, which is in agreement with other studies.33,34 Other authors also emphasized the negative effect of smoking on survival in young patients with head and neck cancer.35 Tobacco consumption seems to induce the tumor cells of oral squamous cell carcinomas to undergo a more pronounced dedifferentiation that makes them more aggressive.36 Therefore, our results suggest that smoking and alcohol cessation programs should be warranted not only with a prophylactic purpose but also with a therapeutic purpose.33
Sex was also identified as a prognostic factor in this study. A lower risk of death was found among women. The survival advantage experienced by females, independent of the effect of other clinical factors, had been already reported in other series.37,38
Malnutrition has been recognized for many years as a comorbid condition in patients with cancer,39 and it is reported to affect 30% to 50% of all patients with head and neck cancer.40 Although evaluation of nutritional status in this study was limited to body mass index and weight loss before treatment, the results are in concordance with literature40,41 that described the negative effect of a poor nutritional status on survival. Immune depression, represented in this study by the total lymphocyte count, is recognized as a consistent metabolic effect in oral cancer, and monitoring of the immunoregulatory status has been shown to be correlated with prognosis.42 It is reported that the preoperative level of malnutrition is associated significantly with postoperative complications and death.43 Therefore, it seems likely that an improvement in nutritional status before surgery, particularly in elderly patients, might decrease postoperative morbidity rates.43
The ASA classification represents a simple estimation of physiologic status without the need for clinical resources, and it can be applied to every patient before surgery.44 The correlation between ASA classification and postoperative morbidity and mortality rates has been shown in previous studies.44,45 The association between ASA classification and prognosis noted in our study is another indication of the effect of poor physical status on outcome. A recent study46 of patients with head and neck cancer concluded that ASA class is comparable to the Charlson index, displaying equal if not greater prognostic value for mortality. This ability was still observed beyond the perioperative period.46
We could not confirm the prognostic value of age found in our previous study,13 probably because of the low number of elderly patients with oropharyngeal carcinomas in this sample. Because surgery carries the risk of aspiration and postoperative complications, there is a tendency to refer these patients to exclusive radiotherapy.
In addition to symptoms, the presence of comorbidities has been shown to decrease survival rates in several chronic diseases,47 demonstrating higher prognostic impact than tumor size or stage in many cancers.48 Moreover, there are preventive implications because alcohol abuse and smoking are major risk factors in head and neck cancer and are also related to other chronic diseases. Primary prevention that begins in an early age and continues throughout life is critical to reduction of the burden of these illnesses.48
Perioperative complications are defined as unexpected but avoidable events that arise during surgery or in the postoperative period.49,50 Complications after major surgery for patients with oral cancer increase treatment costs, delay adjuvant treatment, augment late sequelae, affect quality of life, and can also cause the patient's death if not promptly diagnosed and treated.51 During the past few years, several advances in medical knowledge and techniques improved the safety of major head and neck oncologic surgery, decreasing the risk and severity of complications.52 However, the rates of complications are still high, and the identification of associated risk factors can reduce morbidity and mortality rates in patients with oral or oropharyngeal cancer.53- 55
In our study, the incidence of perioperative complications was similar to that in other studies in the literature.52- 56
Wound infection was the most frequently occurring of all local and systemic complications, as already described in the literature.3,21,55,57 Many recent studies58- 60 have addressed the issue of identifying patients at high risk for developing a wound infection after head and neck oncologic surgery. Advanced stage of tumor,21 type of reconstruction,21,61 preoperative radiotherapy,53 nutritional status,21 comorbidities,21,61 duration of surgery,21 classification of the procedure,21 antibiotic prophylaxis,21,61 tobacco use,61 and alcohol consumption21 have been significantly related to postoperative wound infection. However, the prognostic significance of postoperative wound infection on head and neck cancer remains controversial.62
Chylous fistula is an uncommon complication after neck dissection, occurring in 1.0% to 2.5% of radical neck dissections,63 and the incidence in this study is similar to that in other series.53,64- 66 Our incidence of hematoma is higher than the rates reported by other researchers,53,67 but it is lower than the 4.2% rate described by Johnson and Cummings.68
In the present series, a low rate of pneumonia was detected, whereas in other studies53,69,70 it ranged from 7% to 15%. Although head and neck surgery rarely approaches the pleural space, tracheostomy, pharyngolaryngeal tumor resection, and the use of regional reconstructive flaps from the chest and abdomen may indirectly impact and impair respiratory function.70 The etiology of pulmonary infections after head and neck surgery is most likely multifactorial. Factors may include prolonged anesthesia, resulting in alveolar hypoventilation and atelectasis, as well as aspiration of oropharyngeal secretions during and immediately after surgery.69 The intensive pulmonary care and the low number of patients submitted to hemiglossectomy or total glossectomy with risk for aspiration are probably responsible for this low rate of pneumonia in our patients.3
The mortality rate (2.6%) in our study is comparable to that in other studies.52,53 The causes of postoperative death in our study were arterial rupture and hypovolemic shock, respiratory distress, bronchopneumonia, acute pulmonary edema, pulmonary embolism, cardiac arrhythmia, and sudden death.
The choice of a neck dissection is based on the primary site and on the number, size, and location of positive lymph nodes. In addition, the results and morbidity associated with each type of neck dissection must be considered. Bilateral radical neck dissection has been used for many years for the treatment of proved or suspected bilateral metastatic disease, carrying significant morbidity and mortality rates.71
The use of microvascular-free tissue transfer has warranted the reconstruction of increasingly complex defects in high-risk patients after head and neck cancer surgical treatment.72 However the association of these factors also originates a higher risk for complications,3,72 as occurred in our study.
The prognostic value of the APACHE II has been shown in patients having major general surgery, with indications that included colorectal cancer, breast cancer, gastric cancer, ovarian cancer, lymphoma, and pancreatic cancer.73 In this study, we confirmed our previous finding3 that APACHE II score predicts perioperative complications. To our knowledge, this is first study to establish the prognostic value of the APACHE II score for patients with head and neck cancer.
Clinical staging systems that include patients' clinical characteristics have been developed and have shown prognostic gradients for several types of cancer. Although the customary staging systems describe the morphologic appearance of and the structural damage produced by the tumor, no attention is given to the tumor's duration and rate of growth, which can be revealed as the functional effects of the cancer in structures or systems that may or may not be anatomically involved.74,75
The ECSS system developed in this study provides a better estimate of 5-year overall survival, tumor-specific survival, and recurrence rates compared with the TNM. The superiority of this kind of staging system was previously reported in various other studies focusing on oral,12,13 oropharyngeal,11 or laryngeal cancer,9 confirming the hypothesis that clinical variables have important prognostic value.
Survival estimates in head and neck cancer can be improved by the addition of clinical elements to the TNM classification, creating a more powerful and precise staging system. Prediction of outcome is important in disease stratification and subsequent decision-making processes.73 Identification of risk factors for perioperative complications may help the surgeon classify patients into groups with distinct probabilities of postoperative morbidity and mortality. Thus, the reduction of risk factors for perioperative complications and chronic diseases can turn into a better prognosis in patients with oral or oropharyngeal cancer.
Corresponding author and reprints: Luiz Paulo Kowalski, MD, PhD, Department of Head and Neck Surgery and Otorhinolaryngology, Centro de Tratamento e Pesquisa Hospital do Câncer A. C. Camargo, R. Professor Antônio Pudente, 211, CEP 01509-010 São Paulo-SP, Brazil (e-mail: firstname.lastname@example.org).
Accepted for publication July 18, 2002.