Time to diagnosis in esophageal cancer: a cohort study

Abstract

Background

The association between shorter time to diagnosis and favorable outcome is still unproven in esophageal cancer. This study aims to evaluate the effect of time to diagnosis on patient prognosis.

Material and methods

Retrospective cohort study of all 3613 symptomatic patients referred for esophageal cancer to our center from 1980 to 2011. Time to diagnosis was calculated as the number of days from first symptom onset to the diagnosis of esophageal cancer. The main outcome measures were: resectability and severe malnutrition at diagnosis; postoperative morbidity, mortality and survival.

Results

Longer time to diagnosis was significantly associated with severe malnutrition at diagnosis (odds ratio (OR): 1.003, 95% confidence interval (C.I.).: 1.001–1.006) but not with resectability (OR: 0.997, 95% C.I.: 0.994–1.001). Longer time to diagnosis was not associated with postoperative morbidity (OR: 1.000, 95% C.I.: 0.998–1.003), postoperative mortality (OR: 1.002, 95% C.I.: 0.998–1.006), five-year overall survival (hazard ratio (HR): 0.999, 95% C.I.: 0.997–1.001) or five-year disease free survival (HR: 0.999, 95% C.I.: 0.998–1.001).

Conclusion

Longer time to diagnosis did not affect resectability, postoperative morbidity or survival. Further campaigns to raise awareness of cancer among population and primary health care providers may have limited effect on clinical outcome.

Introduction

Esophageal cancer is the sixth leading cause of cancer death in the world, due to several reasons including ineffective screening tools and advanced disease at diagnosis [1]. Promptness of diagnosis is a key point in cancer control policies, in order to reduce the incidence of late stage at diagnosis and to improve therapeutic efficacy [2–4]. In symptomatic cancer patients, appropriate interpretation of symptoms plays an important role in timely diagnosis [5], but common delays in presentation lead to delays in establishing the diagnosis [6]. The main reason is an underestimation of the initial symptoms, which may not be recognized as serious or may be attributed to comorbidities [7,8].

Although later stage at diagnosis is one of the most important risk factor to poor cancer outcome [9], the association between shorter time to diagnosis and favorable outcome is unproven. A recent systematic review concluded that prompt diagnosis was likely to have benefits for cancer patients, but there was a considerable variation in the amount and in the consensus of these benefits among different type of cancers [10]. Most esophageal cancer patients report a long interval between symptoms onset and presentation (or diagnosis) [6,11], but there is insufficient evidence regarding the association between diagnostic intervals and stage at diagnosis [10], thus additional campaigns to raise awareness of esophageal cancer may lead to over-investigation [12].

The aim of this study was to evaluate the effect of longer time to diagnosis on patient status at diagnosis and long-term prognosis. In addition, the risk factors of longer time to diagnosis were investigated.

Material and methods

Study design

This study was performed following the suggestions of the Aarhus checklist [13], in order to improve precision and transparency in reporting diagnostic intervals. It was conducted according to Helsinki Declaration principles and patients gave their consent to have their data collected for scientific purpose. The study was also notified to the Research Ethical Committee of Veneto Oncology Institute IOV IRCCS that did not find any ethical problems (2014-06-16-Note5).

Patients

All 4440 patients referred for primary esophageal squamous cell carcinoma or adenocarcinoma to the Centre for Esophageal Diseases located in Padova between January 1 1980 and December 312011, were retrospectively evaluated. Asymptomatic patients, those undergoing regular endoscopic screening for Barrett esophagus, those undergoing endoscopic follow-up for previous upper gastrointestinal neoplasm and those not recalling the exact moment of symptom onset) were excluded from the main analysis (Figure 1).

Figure 1. Flow chart of patient selection.

Variables

Patient characteristics, tumor features and follow-up data were extracted from a prospectively maintained electronic database [14]. The time interval of interest was the time to diagnosis (TTD) was defined as the number of days from first symptom onset to the diagnosis of esophageal cancer. The date of first symptom was defined as the date when the first bodily change was noticed by the patient. The date of diagnosis was defined as the date of first histological confirmation of the malignancy. These data and definitions were chosen according to the Aarhus checklist, which is a resource for early cancer-diagnosis research aiming at promoting greater precision and transparency in definitions and methods [13]. At our Centre, information about symptom onset is usually directly asked to the patient during the first interview at outpatient clinic and is recorded in the electronic database, as previously suggested [10,13].

Primary outcomes were conditions of resectability and severe malnutrition (measured as weight loss >10%) at diagnosis. Weight loss was calculated using actual weight at the time of diagnosis and patient-reported ‘normal’ weight. Patients were considered not potentially resectable if they had clinical T4 or M1 classification (seventh edition of AJCC cancer staging system [14]) at diagnosis and resectable otherwise. We included weight loss among nutritional outcomes rather than symptoms because esophageal cancer involves alimentary difficulties (i.e., dysphagia and vomiting) that are usually responsible for weight loss. Secondary outcomes were postoperative morbidity and mortality (within 30 days from surgery), five-year overall survival (OS) and five-year disease free survival. In addition, the risk factors of longer TTD were investigated among potentially relevant clinical and demographic characteristics.

Healthcare context and tumor staging

The Centre for Esophageal Diseases in Padova is a tertiary referral center for patients with esophageal cancer in Italy. Most patients are directly referred from the oncologist of referring hospitals whereas a minority of patients are referred to our outpatient clinic for multidisciplinary oncological consultation about treatment after the endoscopic diagnosis has been established by the gastroenterologist. In all patients, upper gastrointestinal endoscopy with biopsy is (re)done in our center to confirm the diagnosis of esophageal cancer and to determine the exact location of the tumor. Pre-treatment evaluations for esophageal cancer included barium tests; esophageal endoscopy; computed tomography (CT) of the neck, chest and abdomen; and bronchoscopy. Endoscopic ultrasonography (EUS) of the esophagus has been part of our routine esophageal cancer staging since 2000 and positron emission tomography scan was included since 2005. Every patient is discussed in a weekly multidisciplinary oncology meeting in which a definitive treatment plan is designed. If eligible for surgery, patients are put on the waiting list for surgery. If needed, additional cardiac and/or pulmonary function tests are scheduled.

The surgical treatment consisted in open radical transthoracic esophagectomy with cervical or mediastinal anastomosis. Details concerning surgical techniques have been published elsewhere [15]. Palliative treatment was performed on unresectable neoplasms or patients unfit for surgery. It included nose-gastric tube placement, endoscopic dilatation, laser ablation, but more often prosthesis placement and surgical by-pass procedures in selected patients. Follow-up visits were scheduled every three months in the first year after surgery, every six months during the next two years and every 12 months thereafter. An upper gastrointestinal endoscopy was performed regularly one year after surgery or earlier based on the clinical findings, with direct evaluation of the remaining esophagus, anastomosis and of the esophageal replacement conduct. Functional results were assessed based on clinical and endoscopic findings.

Statistical analysis

Continuous data were expressed as median and interquartile range (IQR), categorical data as number of patients and percentage. Missing data were due to the retrospective nature of the study and were not imputed.

The effects of TTD on resectability was estimated with a logistic regression model, adjusting for clinically relevant confounders (age, sex, year of diagnosis, histotype and tumor site). Clinical stage was not included among confounders for resectability, because resectability was defined according to clinical stage: patients were considered not potentially resectable if they had clinical T4 or M1 classification [14] at diagnosis and resectable otherwise. The effects of TTD on severe malnutrition was estimated with a logistic regression model, adjusting for clinically relevant confounders (age, sex, year of diagnosis, histotype, tumor site and clinical stage).

In resected patients, the effects of TTD on postoperative morbidity and on postoperative mortality were estimated with logistic regression models, adjusting for clinically relevant confounders (period of diagnosis, age, malnutrition, tumor location and histotype). The effect of TTD on five-year OS in all patients was estimated with a Cox regression model, adjusting for clinically relevant confounders (period of diagnosis, age, sex, histotype, tumor site, neoadjuvant therapy, surgical resection and tumor, node and metastasis (TNM) stage). The effect of TTD on five-year disease free survival (DFS) in resected patients was estimated with a Cox regression model, adjusting for clinically relevant confounders (period of diagnosis, age, sex, histology, tumor site, neoadjuvant therapy and TNM stage). OS and DFS were calculated according to proposed guidelines on cancer endpoints [16].

A Gamma model with log link function was estimated to identify independent predictors of TTD, among potentially relevant clinical and demographic characteristics (calendar period of diagnosis, age, sex, hiatal hernia/reflux disease, histotype, tumor site, tumor lengthand area of residence).

As additional analysis, symptomatic patients were compared with excluded patients (those undergoing regular screening for Barrett esophagus, those undergoing endoscopic follow-up for previous upper gastrointestinal neoplasm, asymptomatic patients and those without information about symptoms onset) in term of resectability, severe malnutrition and five-year OS.

All tests were two-sided. A p value less than .05 was considered statistically significant. Statistical analysis was performed using SAS 9.1 (SAS Institute Inc., Cary, NC, USA).

Results

Patients

The study included 3613 symptomatic esophageal cancer patients (953 esophageal adenocarcinoma and 2660 esophageal squamous cell carcinoma) with a median age of 62 years (IQR 55–69). Median TTD was 90 days (IQR 60–150). Patient characteristics are shown in Table 1.

Table 1. Patient characteristics.

Variables	Number (%) or median (IQR)
Total cohort	3613
Year of diagnosis:
1980–1989	1666 (46.1)
1990–1999	936 (25.9)
2000–2011	1011 (28.0)
Time to diagnosis, days	90 (60–150)^a
Time to diagnosis, weeks	13 (9–21)^a
Age at diagnosis, years	62 (55–69)^a
Sex (male: female)	2938:675
Area of residence:
Northeastern Italy	2727 (61.6)
Northwestern/Central/Southern Italy	1693 (38.4)
Histotype:
Esophageal adenocarcinoma.	953 (26.4)
Esophageal squamous cell carcinoma	2660 (73.6)
Tumor site:^b
Hypopharynx and cervical esophagus^c	419 (11.6)
Upper and middle thoracic esophagus	1714 (47.6)
Lower thoracic esophagus and esophagus-gastric junction	1469 (40.8)
Clinical tumor stage:
I	219 (6.1)
II	1478 (40.9)
III	1407 (38.9)
IV	509 (14.1)
Tumor length, cm	6 (4–8)^a
Neoadjuvant therapy	1338 (37.0)
Surgical treatment:
No surgery/surgical exploration	566 (15.7)
Palliative treatment	989 (27.4)
Surgical resection	2058 (56.9)

^aMedian (IQR).

^bOther site in 11 patients.

^cTumor site was hypopharynx in 92 patients and cervical esophagus in 327 patients.

Primary outcomes

At diagnosis, 1201 patients were not considered resectable due to clinical T4 classification (706 patients), clinical M1 classification (355 patients) or both (140 patients) as shown in Table 2. TTD was not associated with resectability (odds ratio (OR) 0.997 per additional week, 95% confidence interval (C.I.): 0.994–1.001. Longer TTD was an independent predictor of severe malnutrition at diagnosis (OR 1.003 per additional week, 95% C.I.: 1.001–1.006). According to the model, the probability of severe malnutrition increased by 0.3% for each additional week in TTD or by 0.7% for every additional two weeks or by 1.3% for every additional month.

Table 2. Effect of time to diagnosis on primary outcomes.

Primary outcomes	Patients (number)	Odds ratio (95% C.I.)^a	p value
Resectability:^b			.110
No	1201	0.997 (0.994–1.001)
Yes	2412	Reference
Weight loss at diagnosis:^c			.020
≤ 10%	2247	Reference
> 10%	1068	1.003 (1.001–1.006)^d

^aOdds ratio for each additional week in time to diagnosis.

^bAdjusted for age, sex, calendar period of diagnosis, histotype and tumor site. Patients were considered not resectable if they had clinical T4 or M1 classification at diagnosis and resectable otherwise.

^cAdjusted for age, sex, calendar period of diagnosis, histotype, tumor site and clinical stage. Data on weight loss were not available in 298 patients.

^dAccording to the model, the probability of severe malnutrition increased by 0.3% for each additional week in time to diagnosis or by 0.7% for every additional two weeks or by 1.3% for every additional month.

Secondary outcomes

Among 2058 patients who underwent surgical resection, postoperative complications occurred in 935 patients (45.4%) and postoperative mortality was 5.6% (116 patients). TTD was not associated with postoperative morbidity (OR: 1.001 per additional week, 95% C.I.: 0.998–1.004) or postoperative mortality (OR: 1.002 per additional week, 95% C.I.: 0.998–1.006) as shown in Table 3. Median follow-up was 11 months (IQR 5–26 months). OS at five years from diagnosis was 17.1% for the whole sample, with higher rate for patients who underwent esophagectomy (27.2 vs. 3.3% in patients who did not undergo esophagectomy, p < .0001). DFS at five years from surgery was 25.7% among 2058 patients who underwent surgical resection. TTD was not associated with OS (hazard ratio (HR): 0.999 per additional week, 95% C.I.: 0.997–1.001) or DFS (HR: 0.999 per additional week, 95% C.I.: 0.998–1.001) as shown in Table 3.

Table 3. Time to diagnosis and secondary outcomes.

Secondary outcomes	Patients	Patients (number)	Odds ratio (95% C.I.)^a	p value
Postoperative morbidity^b	Resected	2058	1.001 (0.998–1.004)	.608
Postoperative mortality^b	Resected	2058	1.002 (0.998–1.006)	.260
			Hazard ratio (95% C.I.)^a	p value
5-year overalls survival^c	All	3613	0.999 (0.997–1.001)	.382
5-year disease-free survival^d	Resected	2058	0.999 (0.998–1.001)	.598

^aOdds ratio (and hazard ratio) for each additional week in time to diagnosis.

^bAdjusted for calendar period of diagnosis, age, malnutrition, tumor site and histotype.

^cAdjusted for calendar period of diagnosis, age, sex, histology, tumor site, neoadjuvant therapy, surgical resection and tumor stage.

^dAdjusted for calendar period of diagnosis, age, sex, histology, tumor site, neoadjuvant therapy and tumor stage.

Factor associated with longer time to diagnosis

Multivariable analysis (Table 4) identified female (estimated effect 1.23, 95% C.I.: 1.15–1.32) and hiatal hernia/reflux disease (estimated effect 1.58, 95% C.I. 1.39–1.82) as independent predictors of longer TTD. Calendar period of diagnosis 1990–1999 (estimated effect 0.88, 95% C.I.: 0.82–0.94) and 2000–2011 (estimated effect 0.74, 95% C.I.: 0.70–0.79 ), longer tumor (estimated effect 0.99, 95% C.I.: 0.98–0.99 ) and living in North-East Italy (estimated effect 0.94, 95% C.I.: 0.90–0.99) were independent predictors of shorter TTD.

Table 4. Multivariable analysis of time to diagnosis.^a

Variables	Patients (number)	Estimated effect (95% C.I.)	p value
Calendar period of diagnosis:
1980–1989	1666	Reference
1990–1999	936	0.88 (0.82–0.94)	<.0001
2000–2011	1011	0.74 (0.70–0.79)	<.0001
Age at diagnosis:
<70 years	789	Reference
≥70 years	2824	0.99 (0.95–1.08)	.769
Sex:
Male	2938	Reference
Female	675	1.23 (1.15–1.32)	<.0001
Hiatal hernia/reflux disease
No	3453	Reference
Yes	160	1.58 (1.39–1.82)	<.0001
Histotype:
Esophageal adenocarcinoma	953	1.06 (0.98–1.15)	.148
Esophageal squamous cell carcinoma	2660	Reference
Tumor site:^b
Hypopharynx and cervical esophagus	419	1.07 (0.97–1.19 to 1.19)	.170
Upper and middle thoracic esophagus	1714	0.96 (0.89–1.03)	.226
Lower thoracic esophagus and esophagus-gastric junction	1469	Reference
Tumor length, cm^c	3510	0.99 (0.98–0.99)	.006
Area of residence:
Northeastern Italy	2727	0.94 (0.90–0.99)	.043
Northwestern/Central/Southern Italy	1693	Reference

^aMultivariable analysis was performed by Gamma model with log link function.

^bOther site in 11 patients.

^cNot available in 103 patients.

Additional analysis with excluded patients

Asymptomatic patients, those undergoing regular endoscopic screening for Barrett esophagus and those not recalling symptoms onset were more likely to be resectable at tumor diagnosis, were less likely to report severe malnutrition and were associated with improved five-year overall survival with respect to symptomatic patients (Supplementary Table 1). Patients undergoing endoscopic follow-up for previous upper gastrointestinal neoplasm did not show any advantages in resectability, severe malnutrition or five-year survival with respect to symptomatic patients (Supplementary Table 1). All analyses were adjusted for clinically relevant confounders.

Discussion

Available literature shows limited consensus on diagnostic delay in cancer patients and highlights the presence of heterogeneity and unaddressed bias that precludes definitive conclusion of the real-life impact of time to diagnosis on the prognosis of patients [3,10,17].

Analysis of TTD is important in esophageal cancer because most patients report obstructive symptoms suggestive of cancer [18] but TTD remains longer than in other cancers, as renal, breast and lung cancer [6]. Our data showed a median TTD of three months that was in broad agreement with previously reported intervals of 3–4 months for esophageal cancer [19,20]. Such long intervals might be due to the underestimation of symptom seriousness, that has been recognized more important than the recognition of the symptom presence [21,22].

The effectiveness of information campaigns would be related with the identification of patients at risk of longer TTD but, to our knowledge, the risk factors of longer TTD in esophageal cancer have not been established yet. Our multivariable analysis identified living in Northeastern Italy and being diagnosed in 1990–1999 and in 2000–2011 as independent predictors of shorter TTD. Northeastern Italy has the highest incidence of esophageal cancer among Italian geographical areas, thus it may be associated with high awareness and attention of local health providers [23,24]. In addition, global increase in cancer awareness might be associated with the decrease of TTD after the 80s. A longer tumor length was also associated with shorter TTD, but its effect was limited. Multivariable analysis identified female sex and hiatal hernia/reflux disease as independent predictors of longer TTD. Previous studies suggested that women are more likely to underestimate cancer symptoms and also delay in seeking care than men [25], but evidence of such difference is inconclusive [6]. Hiatal hernia and reflux disease are well-known risk factors for esophageal cancer, but they are also common in general population and have symptoms similar to esophageal cancer ones [26,27]. These aspects increase the relevance of the non-recognition of symptom seriousness and lead to longer TTD. Unfortunately, the amount of patient’s delay included in TTD could not be identified in our data.

A recent systematic review has shown the importance of TTD in cancer research but also highlighted the inconclusiveness of studies on esophageal cancer [10]. The review included two studies [19,21] reporting a positive association between short delay interval and early stage at diagnosis and two studies [28,29] reporting no association. In the present study, clinical stage was used to identify patients with resectable or unresectable tumors at diagnosis, because we focused on the chance of cure. Our findings suggested that the possibility of cure (i.e., resectability and survival) was not associated with TTD. Moreover, the results did not evince the presence of the waiting time paradox, according to previous reports [10]. These findings suggest that further campaigns to raise awareness of cancer may have limited effect on clinical outcome.

Symptomatic esophageal cancer patients usually report alimentary difficulties that are responsible for weight loss [16], therefore a longer delay might cause a worse nutritional status [16]. Our findings confirmed this association but the marginal effect of TTD on the probability of severe malnutrition was limited. However, once the diagnosis is made, malnourished patients usually receive nutritional support before surgery in order to reduce the probability of postoperative complications. In fact, in our series, postoperative morbidity and mortality were not related to TTD.

Interestingly, asymptomatic patients and those not recalling symptoms onset showed favorable status at diagnosis and improved survival with respect to symptomatic patient, thus suggesting a less invasive esophageal cancer. In addition, patients undergoing regular endoscopic screening for Barrett esophagus were also associated with favorable status at diagnosis and improved survival. This finding underlines the potential benefits of endoscopic screening for esophageal cancer in high-risk European populations, given the results showed in other countries (i.e., Japan and South Korea) [30].

The strengths of this study rely in methodological aspects. Data collection was described according to the Aarhus statement [13] and information on TTD have been routinely collected by an in-depth qualitative interview of the patients at the outpatient clinic [10,13]. Moreover, TTD have been evaluated as continuous variable rather than a dichotomous variable, in order to avoid an underestimation of the variability [31].

The present study has also some limitations. First, it is a retrospective study of single-center experience. Second, the sample included esophageal cancer patients referred to a third level center, thus the results are not completely generalizable to all esophageal cancer patients. Third, some points of Aarhus checklist [13] could not be filled due to the retrospective nature of the study. In addition, information on interim intervals (i.e., time to first presentation and primary care interval) were not available, therefore the effect of patient delay, primary care delay and the effect of seasonal holidays on delayed start of treatment after the first histological diagnosis could not be assessed. Finally, the study enclosed a long time span, but all analyses were adjusted for the period of diagnosis.

In conclusion, longer TTD in esophageal cancer was associated with severe malnutrition at diagnosis but did not affect resectability and both short- and long-term outcomes. Further campaigns to raise awareness of cancer among population and primary health care providers may have limited effects on clinical outcome. Endoscopic screening for esophageal cancer in high-risk populations may increase endoscopic treatments at reduced costs. Additional studies are required to identify specific advantages and disadvantages of awareness campaigns and to assess cost-benefit analysis of endoscopic screening.

Disclosure statement

The authors declare that they have no competing interests.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.