Kaplan-Meier plot of survival for subjects with presentation-to-diagnosis interval (PDI) less than 3 months and 3 months or longer.
Ho T, Zahurak M, Koch WM. Prognostic Significance of Presentation-to-Diagnosis Interval in Patients With Oropharyngeal Carcinoma. Arch Otolaryngol Head Neck Surg. 2004;130(1):45-51. doi:10.1001/archotol.130.1.45
To evaluate the prognostic impact of presentation-to-diagnosis interval (PDI) and its association with other clinical factors in patients with oropharyngeal squamous cell carcinoma (OpSCC).
Retrospective cohort study.
Otolaryngology clinic of an academic medical center.
Eighty-seven patients with OpSCC referred to the otolaryngology service at the Johns Hopkins Medical Institutions from March 1994 to August 2001 were included in the study. Selection criteria included confirmed pathological diagnosis of OpSCC, availability of referral record for PDI determination, and no past history of oropharyngeal cancer.
Main Outcome Measures
The PDI is defined as the time between the patient's first presentation to a medical professional for tumor-related symptoms and the time when the diagnosis of OpSCC was made. The prognostic impact of PDI and its association with other clinical factors were assessed using univariate and multivariate analyses.
Forty percent of patients (35/87) had a PDI of 3 months or longer. Referred otalgia, active smoking status at the time of diagnosis, stage IV disease, and advanced T stage were associated with a poor prognosis. Prolonged PDI itself was not associated with a significant decrease in survival in univariate analysis (hazard ratio, 1.27; P = .52). Furthermore, no significant correlation was found between PDI and N stage, T stage, young age at presentation (<45 years), or tobacco use.
Difficulty in making the diagnosis of OpSCC is evident by the high proportion of patients with PDI of 3 months or longer. The PDI does not appear to have an impact on survival. Referred otalgia, widely recognized as a strong indicator of invasive head and neck cancer, portends a poor prognosis.
More than 90% of the malignant tumors found in the oropharynx are squamous cell carcinomas (OpSCC). The tonsils (lingual and palatine) and faucial pillars are the primary tumor sites in approximately half of all cases.1,2 The malignancy usually presents at an advanced stage when the tumor has attained at least a T2 classification (>2.0 cm) and in 60% of the cases there is evidence of regional lymphatic channel involvement attributed to the rich lymphatic supply in this anatomic region.3 The disease is associated with high mortality, and treatment brings substantial morbidity to this functionally important anatomic region. Despite advances in treatment options, the prognosis has remained largely unchanged for the past 3 decades.
There is a significant correlation between prognosis and the stage of disease at presentation, making early diagnosis crucial.4 Much of the oropharynx can be inspected and palpated without special instrumentation.2 In theory, this provides an opportunity for early diagnosis of OpSCC, yet in reality this can be difficult. Some of the factors that have been cited to contribute to the late stage of diagnosis include inadequate access to routine medical care; the irregular anatomic surface of the oropharynx, making inspection difficult; and the lack of alarming early symptoms, resulting in a lack of impetus to seek medical attention.5 In previous studies a diagnostic interval of 12 to 17 weeks was attributed to patient factors. In addition, 30% of the cases had professional factors contributing to the prolonged diagnostic interval.3,6 To facilitate achieving the diagnosis at an earlier stage, we need to have a better understanding of the factors that influence the presentation-to-diagnosis interval (PDI).
Progress has been made on understanding the progression of OpSCC on a molecular level. In the last decade, the critical genetic mutations that are key in the sequence of events that lead to the growth of invasive head and neck cancer from preneoplastic lesions have been described.1 Recently it was found that more than 50% of oropharyngeal carcinomas contain oncogenic human papillomavirus (HPV) DNA.7 Patients with tumors that were HPV positive were also found to be less likely to harbor p53 mutations and had better overall survival as well as disease-specific survival compared with patients with HPV-negative tumors.8
Some studies on early diagnosis of head and neck malignancies have shown conflicting results regarding the clinical impact of the time lapsed from when patients first presented to their health care providers until diagnosis. These discrepancies often are largely attributed to the different methods used in collecting data and the various definitions of delay. There have been a few studies examining the impact of both professional and patient attitudes and actions on the timely diagnosis and prognosis of upper aerodigestive tract cancers. We are also not aware of published studies focusing specifically on OpSCC in this regard. The goal of this study was to explore the various factors that may have an impact on the stage of disease at presentation and survival, especially PDI. Ultimately, we hope to increase awareness of the difficulties in diagnosing this malignancy with the intention of promoting earlier diagnosis and treatment.
A comprehensive retrospective analysis of hospital and referral records of a consecutive sample of 124 patients diagnosed with carcinoma of the oropharynx (International Classification of Diseases, Ninth Revision code 146) was performed. All patients were seen at the Johns Hopkins Medical Institutions from March 1994 to August 2001 and were treated with aggressive therapy with curative intent. Treatment plans depended on the location of the tumor and the stage of disease at presentation and consisted of combinations of surgical excision, radiotherapy, and/or chemotherapy. Medical records were obtained from the departments of otolaryngology–head and neck surgery, radiation oncology, and medical oncology. Twenty-five patients were excluded from the study because of incomplete or insufficient referral records or histological types other than squamous cell carcinoma (eg, lymphoma, adenoid cystic carcinoma). An additional 12 patients were excluded because the malignancies were recurrent at the time of referral to our department. This complicates the process of determining PDI in these patients because the diagnosis of recurrent OpSCC was reached in the course of routine cancer surveillance examinations for previous primary tumors.
The PDI was defined as the time between initial medical evaluation by any health care professional for tumor-related symptoms and when a definitive histological diagnosis of squamous cell carcinoma was made. During the review of hospital medical and outside referral records the following variables were recorded: (1) duration of symptoms prior to initial presentation, (2) type of symptoms (neck mass, dysphagia, odynophagia, referred otalgia), (3) date of first presentation to any health care provider with tumor-related symptoms, (4) date of referral to an otolaryngologist, (5) date of referral to Johns Hopkins Medical Institutions, (6) date of biopsy, (7) initial pathological staging, (8) epicenter, (9) risk factors (including prior or family history of malignancy), (10) date and type of surgery (if any), (11) length and type of adjuvant therapy including radiation and chemotherapy (if any), (12) date of last visit or date of death, and (13) medical interventions administered during PDI.
The major statistical end point of this study was PDI. Factors associated with PDI of 3 months or longer were identified based on cross tabulations and logistic regression modeling. Cross tabulations were analyzed using χ2 or Fisher exact tests where appropriate. Univariate logistic regression models were used to determine the independent effects of multiple factors on PDI.
A second major statistical end point of this study was survival. Event time distributions for this end point were estimated with the method of Kaplan and Meier and compared using the log-rank statistic or the proportional hazards regression model. The simultaneous effect of 2 or more factors was studied using the multivariate proportional hazards model.
All statistical computations were performed using SAS (SAS Institute, Cary, NC) or EGRET (Statistics and Epidemiologic Research Corp, Seattle, Wash) PC packages. All confidence intervals (CIs) are at the 95% level and all P values are 2-sided.
Patient demographics at the time of diagnosis are shown in Table 1. Most patients were white. The male-female ratio of our patient population was approximately 3.15:1. Only a small minority of the subjects were younger than 45 years, and nearly half of the subjects were older than 60 years. The mean age of the entire sample was 57.4 years (range, 37-80 years). The mean ages at the time of diagnosis for men and women were comparable (57.0 and 57.6 years, respectively). The specific risk factors documented included history of smoking, alcohol use, family history of malignancy of any type, and personal history of chemical exposure. More than half of the patients (51%) had 2 of these risk factors documented. We further categorized patients' smoking history into never-smokers, past smokers (abstinence from tobacco use for ≥10 years), and current/recent smokers. More than half of all patients (52%) were current or recent smokers at the time of diagnosis. A majority of the patients initially presented to their health care provider with the chief complaint of palpable neck mass (57%), followed in frequency by pain directly attributed to the oropharynx including dysphagia and odynophagia (38%), while few had referred otalgia (6%).
The primary tumor site was found most commonly in the palatine tonsils (53%) and base of the tongue (33%), which together accounted for more than 80% of the subjects in this series. A small minority of the patients had tumor epicenter located in the soft palate (5%). For the purpose of this study we grouped the posterior and lateral pharyngeal wall as well as the tonsilar pillars together as other sites (10%).
Clinical staging of the tumor using the 1999 American Joint Committee on Cancer TNM staging system at the time of initial presentation is presented in Table 2. A significant majority (85%) of the patients presented in advanced stages (overall clinical stage III or IV). Slightly more than a quarter (28%) of the patients presented with advanced T3 or T4 stage lesions. Also, more than half (79%) of the patients had evidence of nodal metastasis on clinical examination during initial evaluation.
The mean PDI was 3.52 months (range, 0-30 months) for the study population. As shown in Table 3, only 21% of the patients had zero or minimal interval (≤1 week). For the remaining patients the proportion who had a PDI of less than 3 months (39%) was approximately equal to that of those who had a PDI of 3 months or longer (40%). We categorized the patients according to the primary mode of intervention during the PDI (period prior to the diagnosis of OpSCC): 40 (58%) received multiple courses of antimicrobial therapy, 11 (16%) received therapeutic intervention for other presumed diagnoses (eg, antireflux medications, salivary gland stimulants, denture adjustments), 10 (14%) had persistently negative workup by 1 or multiple medical professionals, 8 (12%) had apparent delay in scheduling follow-up or diagnostic testing appointments, and in 15 patients (22%) we were unable to determine the primary course of medical intervention.
Table 4 lists the correlation between PDI and selected clinical and pathological variables. None of the variables examined had statistically significant correlation with PDI. However, there were notable findings. First, patients younger than 45 years had an apparent increased risk (odds ratio [OR], 4.17; P = .10) for experiencing prolonged PDI. Neither sex nor race was noted to have a significant association with PDI. Patients presenting with referred otalgia were observed to have an increased risk for extended PDI (OR, 4.89; P = .18) compared with patients presenting with neck mass, though the association was not statistically significant. Smoking status, epicenter location, and the T and N stage of tumor at the time of initial diagnosis failed to show a significant correlation with PDI. An overall clinical stage IV had an apparent 2-fold increase in risk for an extended PDI (OR, 2.11; P = .13), but again the association was not statistically significant.
Overall 4-year survival was 57%. The median follow-up of surviving patients was 38 months. Table 5 shows the univariate hazard ratio estimate for survival by both clinical and pathological characteristics. Factors increasing the risk of death included current smoking, nonwhite race, overall stage IV disease, soft palate involvement, T3 or T4 disease at time of diagnosis, and referred otalgia as the presenting symptom (as compared with neck mass). White patients had nearly 3-fold reduction in risk (hazard ratio [HR], 0.32; P = .003). In contrast, patients presenting with referred otalgia as the chief complaint had nearly 5-fold increase in risk of death compared with patients presenting with neck mass (HR, 4.61; P = .01). Current smokers also had increased risk of death 3 times that of former smokers or nonsmokers (HR, 3.19; P = .01). Patients with tumor epicenter in the soft palate also had increased risk of death compared with patients with tumor epicenter in the tonsils or the base of the tongue (HR, 2.63; P = .03). Furthermore, patients with advanced T-stage disease also had increased risk of death (HR, 2.72, P = .01). A trend toward increased risk was observed for clinical stage IV disease, although it did not reach statistical significance.
Multivariate results are presented in Table 6. The strongest factors for survival included race, current smoking status, and overall clinical stage. Adjusting for age, race, and current smoking status, patients with stage IV disease had 3.6 times the risk of death compared with patients with stage 0 to III disease (P = .01). Nonwhite race (HR, 3.7; P = .002) and current smoking status (HR, 2.9; P = .01) were also independent predictors of survival when adjusting for the other factors in the model.
A PDI of 3 months or longer did not achieve a statistically significant correlation with survival on univariate analysis; Figure 1 shows the Kaplan-Meier plot for patients with PDI less than 3 months and 3 months or longer.
Oropharyngeal carcinoma has considerable medical and social impact for patients secondary to its anatomic location and morbidity associated with the available treatment options. Moreover, there has been minimal improvement in survival despite advancements in available treatment modalities, including extirpative surgical technique, radiotherapy, and chemotherapy. During the past several decades the prognosis for cancers of the breast, colon, and skin have notably improved largely secondary to an increase in public awareness of the diagnosis leading to earlier detection. In contrast, 5-year survival for oropharyngeal carcinoma has remained fairly constant at 40% to 50%, with more than 50% of the patients noted to have regional cervical metastases at the time of diagnosis.9 Given that the best chance for cure lies in establishing the diagnosis prior to regional spread, a timely diagnosis is clearly necessary.
Intuitively it would appear that a delay of any type would lead to diagnosis at more advanced stage. Our analysis suggests that there is a trend for patients with an extended PDI to be diagnosed with stage IV disease at the time of diagnosis (OR, 2.11; P = .13) but the association is not statistically significant. Indeed, there has been controversy in the existing medical literature regarding the correlation between patient/diagnostic delays and clinical stage at presentation. Allison et al10 in a study of 188 patients with upper aerodigestive tract carcinoma found that professional delay of greater than 1 month is predictive of late-stage disease at time of diagnosis and further conclude that a direct relationship exists between delay and stage of disease at diagnosis. However, in an analysis of 336 consecutive patients with oral and oropharyngeal carcinoma in Brazil neither patient nor professional delay was found to be associated with risk of presenting with advanced disease.11 A similar study of 543 patients with oral and oropharyngeal cancer in Israel also did not find a significant correlation between stage and delay.12 A number of similar studies on oral cavity squamous cell carcinoma also failed to reveal an association between delay and clinical stage.13- 15
In this study we found the mean PDI to be 3.52 months, ranging from none or minimal to 30 months. Data reported by previous studies on OpSCC have been within this range.16,17 Although we collected data on the time interval between when patients first noticed the symptoms and when patients first presented to health care professionals, the inherently subjective nature of patients' own account make this information concerning patient delay difficult to interpret and the data were not included in the final analysis.15 In our study, 40% (n = 35) of the patients had a PDI of 3 months or longer. This observation may be partly attributed to the patient referral pattern seen at this tertiary care center, with many seeking a second opinion on treatment options because they had been diagnostic dilemmas and had gone through numerous diagnostic tests before the final diagnosis of malignancy was made. This explanation is supported by the fact that more than 85% of the patients in this series had late-stage disease (overall clinical stage III and IV) at the time of diagnosis.
No statistically significant correlation between PDI and any of the selected clinical and pathological variables was observed in regression analysis. This is interesting since some of publications on head and neck cancer have reported a correlation between delayed diagnosis and demographic variables such as age and gender.15,16 Furthermore, one intuitively would suspect that patients who have evidence of positive nodal metastasis indicated by positive N stage would have a shorter PDI, but this is not supported by our analysis (OR, 1.23; P = .73). We did, however, observe some intriguing trends. One of these is the increased likelihood of observing an extended PDI for patients who are younger than 45 years. This could be explained by relatively low clinical suspicion secondary to the low proportion of head and neck squamous cell carcinoma in young patients, which has been reported to vary between 0.24% and 9% of all cases.18- 20 Our observation of 15% of young patients in this series is nearly identical to the 15.75% reported by Hart et al21 in a study of 127 patients with oral and oropharyngeal squamous cell carcinoma, which translates into nearly 1 of 7 patients. Similar to the data reported by Hart et al, we did not observe a poorer prognosis for younger patients, in contrast to other literature.22 Similarly, nonsmokers are not recognized as being at risk for head and neck cancer. Less than 2 decades ago the use of tobacco or alcohol was considered to be universal in patients with oral cavity or oropharyngeal malignancies.2 Increased understanding of the molecular progression of head and neck cancer and the role of HPV should raise awareness for the risk of this diagnosis in what was previously thought of as low-risk population such as nonsmokers and young individuals.1,8,23
We observed that for patients who initially presented with referred otalgia as the chief complaint there was a trend for increased risk of having an extended PDI (OR, 4.89; P = .18) and a statistically significant risk for decreased survival (HR, 4.61; P = .01) compared with patients who presented with the chief complaint of neck mass. However, this observation is tempered by the small number of patients who presented with referred unilateral or bilateral otalgia as the chief complaint (n = 5, 6%). The extended PDI observed could be explained by other potential (benign) causes of otalgia entertained by the medical professionals at patient's initial presentation. Patients presenting with this chief complaint were often initially managed with empiric treatment options, observing for symptomatic improvement. Indeed, as shown in our data, antimicrobial therapy was empirically used in approximately 60% of the patients who had a nonnegligible PDI.
Previous studies have suggested that clinical stage at initial presentation is the most important prognostic factor independent of other factors such as sex, age, and tumor location.24 Although one would therefore intuitively suspect that an extended PDI would contribute to a poorer survival, our data did not establish a statistically significant correlation between PDI and survival (HR, 1.27; P = .52). The existing literature on this topic has been consistent. In a series of 167 patients with oral squamous cell carcinoma, Wildt et al15 found no correlation between total delay and survival. Also, Koivunen et al17 found that professional delay did not have a negative impact on survival in a sample of 84 patients with pharyngeal cancer.
Our observation of no statistically significant association between PDI and survival should nonetheless be regarded with caution because of the relatively low power of the study (approximately 9%) to detect a difference in survival. The low power of the study is attributable to the small sample size and the minimal observed hazard ratio for risk of death (HR, 1.27) for PDI of 3 months or longer. In order to have 90% power for detecting an HR of 1.27 we would have had to observe 368 deaths in the group of patients with PDI of 3 months or longer, and we observed 14 in this series. Conversely, with 14 deaths there would have to be an increased relative risk of 3.4 to have 90% power for detecting an effect of PDI in the study. Furthermore, our finding of no statistically significant association between PDI and clinical stage would be consistent with the finding of no association between PDI and survival. However, on subgroup analysis for PDI longer than 6 months and 12 months there appeared to be a decrease in hazard ratio for risk of death with the longer PDI. It is conceivable that this could be due to the difference in the biological behavior of these tumors, in that extended PDI is more common in patients with slow-growing and less aggressive tumors and these patients also have a better prognosis.
In this study we also observed that white patients appeared to have a better prognosis. The reasons for this are not entirely clear. Arbes and Slade25 found that in North Carolina, African American residents had greater odds of been diagnosed with oral cancer at a later stage and speculated that this could be secondary to the fact that African Americans are less likely to have dental visits and therefore oral cancer examination. Although one could speculate that there may be racial differences in access to and utilization of cancer prevention and detection services, the lack of data pertaining to socioeconomic and health services utilization by patients in this series prevents us from making such conclusions. Patients who presented with a neck mass had a more favorable prognosis than did those with a chief complaint of referred otalgia. This might be related to the obvious physical abnormality and availability of techniques such as fine-needle aspiration permitting earlier pathological diagnosis by medical professionals. However, we did not observe a shorter PDI in patients who presented with the chief complaint of neck mass.
It is perhaps not surprising that patients who were active smokers at the time of diagnosis had a poorer prognosis (HR, 3.19; P = .01). In addition to tobacco being a well-established risk factor in OpSCC, the continuation of this risk-seeking behavior despite the presence of cancer-related symptoms could reflect inadequate concern for one's general health status. Smoking also increases the risk of other diseases that could limit overall survival. A statistically significant correlation with poor outcome was also observed for tumors in the soft palate (HR, 2.63; P = .03), which could possibly be related to a molecular pathway not involving HPV. Indeed, tumor subsite was not found to be independent of smoking in the multivariate model. This is consistent with the observation that HPV-related tumors arise in lymphoepithelium and are less likely to be smoking related.
The cause for lack of improvement in survival for patients with oropharyngeal carcinoma is likely to be multifactorial, including limited public awareness of the disease and a changing molecular epidemiology of this malignancy. In a national survey it was found that 40% of adults could not identify any signs of oropharyngeal carcinoma and only 13% could identify alcohol abuse as a risk factor.5 Despite advancement in diagnostic options including modern imaging techniques, an emphasis on early detection by health care professionals as well as raising public awareness of the diagnosis are important and beneficial.26 For instance, it is important for health care professionals, both physicians and dentists, to be aware that the traditional signs of OpSCC such as an ulcerating mass that bleeds are typical of advanced-stage disease and that early lesions may be flat, nonindurated, and may not bleed. A timely diagnosis can be achieved first by a careful visual inspection and manual palpation of the oropharynx since firmness of the tissues may be the only initial sign of a malignancy. Ultimately, any further effort to reduce morbidity and mortality from OpSCC may need to emphasize early detection and raising public awareness.
Several important observations can be extracted from the data of this study. The fact that 40% of all patients presenting with OpSCC reported at least a 3-month interval from first presentation of tumor-related symptoms until diagnosis is indicative of the magnitude of the diagnostic challenge inherent in this disease process and the need for a high index of suspicion on the part of physicians to facilitate earlier disease discovery. The sinister implication of referred otalgia as a presenting sign is underscored. However, the limited scope of this study may account for a failure to verify some of the original hypotheses that prompted its initiation. For example, a larger sample set may confirm the association of prolonged PDI with unusual demographic features for OpSCC such as young age and lack of exposure to tobacco. While PDI almost reaches statistically significant association with advanced stage of disease at presentation, potential association of PDI with reduced survival appears to be more remote. A prospective study involving uniform treatment strategies would provide a more uniform data set from which to make comparisons, although the interval from first presentation to diagnosis cannot always be accurately measured or accounted for. It is suspected that the inherent biological variability involved in disease progression would continue to present a confounding parameter that supersedes the best efforts of patients and care providers in many instances.
Corresponding author: Wayne M. Koch, MD, Department of Otolaryngology–Head and Neck Surgery, Johns Hopkins University School of Medicine, Sixth Floor, 601 N Caroline St, Baltimore, MD 21287 (e-mail: firstname.lastname@example.org).
Submitted for publication March 27, 2003; accepted September 2, 2003.
This study was presented at the annual meeting of the American Head and Neck Society; May 4, 2003; Nashville, Tenn.