Pretreatment Dysphagia in Esophageal Cancer Patients May Eliminate the Need for Staging by Endoscopic Ultrasonography

Abstract

Background

Neoadjuvant therapy is typically given to patients with localized disease who have T3-4 esophageal disease by endoscopic ultrasound (EUS). Previously, we noted that patients who present with dysphagia have a higher EUS T stage. We hypothesized that the presence of dysphagia is predictive of EUS T3-4 disease and that staging EUS could be eliminated for esophageal cancer patients with dysphagia.

Methods

We performed a prospective, intent-to-treat, single-cohort study in which patients with potentially resectable esophageal cancer completed a standardized 4-tier dysphagia score survey. EUS was performed as part of our standard evaluation. To determine whether presence of dysphagia predicted EUS T3-4 disease, the dysphagia score was compared with EUS T stage.

Results

A total of 114 consecutive patients were enrolled between August 2012 and February 2014: 77% (88/114) received neoadjuvant therapy, 18% (20/114) did not, and 5% (6/114) pursued treatment elsewhere. In total, 70% (80/114) underwent esophagectomy. Fifty-four percent (61/114) had dysphagia; 46% (53/114) did not. Dysphagia scores were as follows: 66% (40/61) grade 1, 25% (15/61) grade 2, and 10% (6/61) grade 3-4. Among patients with dysphagia, 89% (54/61) had T3-4 disease by EUS; among those without dysphagia, only 53% (28/53) had T3-4 disease by EUS (p<0.001).

Conclusions

The presence of dysphagia in patients with esophageal cancer was highly predictive of T3-4 disease by EUS. On the basis of this finding, approximately 50% of patients currently undergoing staging EUS could potentially forgo EUS before neoadjuvant therapy. In contrast, patients without dysphagia should still undergo EUS.

INTRODUCTION

The accurate clinical staging of esophageal cancer is a critical component of the treatment for the disease. Systemic, nonsurgical disease is typically identified using computed tomography and positron emission tomography. In patients with local-regional disease, however, endoscopic ultrasound (EUS) is the preferred clinical staging study. EUS is the most accurate means to assess T stage and the current National Comprehensive Cancer Network guidelines recommend EUS for staging of all potentially resectable patients with esophageal cancer (www.nccn.org) [1, 2]. The importance of EUS T staging was established in a recent publication of the CROSS trial. This trial supported that preoperative chemoradiotherapy should be the new standard of care for the management of patients with EUS T3-4 or N1 tumors. Five-year overall survival was 47% for patients treated with neoadjuvant chemoradiotherapy compared with 34% for patients treated with surgery alone [3]. Importantly, the patients in the CROSS trial were all staged with EUS before enrollment.

Many clinicians have known anecdotally that patients who present with dysphagia frequently have an obstructive esophageal tumor that will be staged as T3-4 by EUS. These same patients typically present with a level of dysphagia that can result in malnutrition and dehydration if treatment is delayed for mandatory EUS. We have previously shown, in a retrospective analysis, that the majority of patients who present with dysphagia have a higher T stage [4]. This observation led to the question of whether EUS is necessary for patients who present with dysphagia or whether they can be treated with neoadjuvant therapy on the basis of the presence of dysphagia alone. We hypothesized that patients who present with dysphagia have EUS T3-4 disease and that the presence of dysphagia may therefore eliminate the need for EUS.

Patients and Methods

In this intent-to-treat, prospective trial, we accrued nonrandomized, sequential patients with esophageal cancer. Inclusion criteria included a diagnosis of esophageal cancer with the potential for resection determined at the initial office visit. On the basis of intent-to-treat analysis, all patients were included regardless of whether they ultimately underwent resection. Exclusion criteria included a histologic diagnosis other than adenocarcinoma or squamous cell carcinoma.

At the initial consultation, all patients were evaluated for symptoms of dysphagia by a research assistant who was blinded to clinical test results. The symptoms of dysphagia were recorded on the basis of a published, standardized 4-tier dysphagia score survey [5]. Dysphagia scores are as follows: Able to eat a normal diet (0), able to swallow some solid foods (1), able to eat semisolids only (2), able to swallow liquids only (3), and unable to swallow (4). EUS was performed as part of our standard evaluation to determine T stage in all patients with potentially resectable esophageal cancer before neoadjuvant therapy. The association between the presence of dysphagia (any grade) and EUS T stage (3-4 versus 1-2) was tested using Fisher's exact test to determine whether the presence of dysphagia predicted EUS T3-4 disease. P values were 2-tailed and p<0.05 was considered to indicate statistical significance.

All patients gave informed consent to participate in the study. The study was approved by the Institutional Review Board at Memorial Sloan Kettering Cancer Center.

Results

In total, 114 consecutive patients were enrolled between August 2012 and February 2014. Of the patients, 82% (93/114) were men, most had comorbidities, and 21% (24/114) had had a previous malignancy. Demographic details are listed in Table 1. Ninety-six percent (110/114) had adenocarcinoma; the remaining 4% (4/114) had squamous cell carcinoma.

Table 1.

Patient Characteristics (N=114)

Characteristic	No. (%)
Female sex	21 (18)
Age at enrollment, years, median ± SD	63 ± 10
Age at operation, years, median ± SD	63 ± 10
Comorbiditya
None	16 (14)
Pulmonary	25 (22)
Cardiac	85 (75)
NIDDM	22 (19)
Barrett's esophagitis	19 (17)
GERD	53 (46)
Neurological/psychiatric	28 (25)
Prior malignancyb
Total	24 ( 21)
Leukemia/lymphoma	2 (2)
Colon cancer	3 (3)
Basal/squamous cell skin cancer	10 (9)
Renal cell carcinoma	2 (2)
Melanoma	4 (4)
Other cancers	5 (4)

GERD, gastroesophageal reflux disease; NIDDM; non-insulin-dependent diabetes mellitus.

^aTotal is >100% because some patients had >1 comorbidity.

^bTotal is >21% because some patients had >1 cancer.

Seventy-seven percent of patients (88/114) received neoadjuvant therapy, 18% (20/114) underwent surgery only, and 5% (6/114) pursued treatment elsewhere. Ninety-four percent (83/88) received chemoradiotherapy; the remaining 5 patients received chemotherapy only. The doublet of carboplatin and paclitaxel (81%; 71/88) was the most common regimen administered, and most patients received 5040 cGy of radiotherapy (83%; 62/75). Details of induction therapy are listed in Table 2.

Table 2.

Details of Induction Therapy

Variable	No. (%)
Induction therapy
None	20/114 (18)
No consent/treatment elsewhere	6/114 (5)
Total induction therapy	88/114 (77)
Chemotherapy followed by CRT	57/88 (65)
CRT	26/88 (30)
Chemotherapy only	5/88 (6)
Radiotherapy dosea
Total known dose	75/83 (90)
Unknown	8/83 (10)
5040 Gy	62/75 (83)
Other	13/75 (17)
Chemotherapy regimenb
Carboplatin/paclitaxel	71/88 (81)
5FU/oxaliplatin	8/88 (9)
Other	5/88 (6)
Unknown	4/88 (5)

5FU, fluorouracil; CRT, chemoradiotherapy.

^a83 patients received radiotherapy.

^b88 patients received chemotherapy.

Seventy percent of patients (80/114) underwent esophagectomy; 54% of esophagectomies (43/80) were open approaches, and 46% (37/80) were minimally invasive approaches. Most of the open and minimally invasive approaches were Ivor Lewis esophagectomies. Ninety-five percent of patients (76/80) had R0 resections. The mean number of lymph nodes removed was 24. One patient died within 90 days of surgery (myocardial infarction). Operative details are listed in Table 3. Thirty-two patients did not undergo resection, for the following reasons: Metastatic disease (n=5), disease progression (n=10), poor performance status or death after induction therapy (n=5), treatment elsewhere (n=6), other reason (n=4), and endoscopic mucosal resection (n=1). Two patients underwent gastrectomy for stomach cancer.

Table 3.

Operative Details

Detail	No. (%)
Total undergoing esophagectomy	80/114 (70)
Total undergoing gastrectomy	2/114 (2)
Estimated blood loss, median ± SD	324 ± 284
Open esophagectomy	43/80 (54)
Ivor Lewis	40/43 (93)
Thoracoabdominal	2/43 (5)
Colonic interposition	1/43 (2)
Minimally invasive esophagectomy	37/80 (46)
Robotically assisted	35/37 (95)
Ivor Lewis	37/37 (100)
Converted to open	1/37 (3)
Resection margin
R0	76 ( 95)
R1	4 (5)
Lymph node dissection, mean ± SD
Number of metastatic lymph nodes	1.4 ± 2.8
Number of lymph nodes dissected	24 ± 9
Location of tumor
Middle	1 (1)
Lower	71 (89)
Unknown	9 (11)
Siewert type
I	40 (50)
II	17 (21)
III	4 (5)
Stomach	1 (1)
Unknown	18 (23)
Mortalitya
30-day	1 (1)
90-day	1 (1)

^aPostoperative myocardial infarction.

Details of pretreatment clinical staging and postinduction pathological staging are listed in Table 4. Of note, the most common preinduction clinical stage was IIIA (32%; 37/114), but only 9% of patients who underwent esophagectomy (7/80) had pathological stage IIIA disease. Two percent of the study patients (2/114) had clinical stage 0 disease, which increased to 23% (18/80) on pathological staging. All patients successfully had a pre-induction EUS performed to determine T stage and no patients were unable to undergo EUS secondary to obstructing tumors. On preinduction clinical staging by EUS, 74% (84/114) had T3-4 disease, and 23% (26/114) had T1-2 disease; on postinduction pathological staging, 33% (26/80) had T3-4 disease and 44% (35/80) had T1-2 disease. The remaining patients had pathological Tis or T0 disease.

Table 4.

Clinical and Pathological Staging

Variable	No. (%)
Pathological diagnosisa
Adenocarcinoma	110 (96)
Squamous cell carcinoma	4 (4)
Preinduction clinical stagea
0	2 (2)
IA	8 (7)
IB	3 (3)
IIA	4 (4)
IIB	30 (26)
IIIA	37 (32)
IIIB	11 (10)
IIIC	9 (8)
IV	8 (7)
Unknown	2 (2)
Preinduction G stagea
GX	7 (6)
G1	6 (5)
G2	55 (48)
G3	43 (38)
G4	1 (1)
Unknown	2 (2)
Clinical T statusa
Tx	3 (3)
Tis	1 (1)
T1	10 (9)
T2	16 (14)
T3	74 (65)
T4	10 (9)
Clinical N statusa
NX	4 (4)
N0	39 (34)
N1	54 (47)
N2	16 (14)
N3	1 (1)
Postinduction pathological stageb
0	18 (23)
IA	15 (19)
IB	6 (8)
IIA	3 (4)
IIB	18 (23)
IIIA	7 (9)
IIIB	8 (10)
IIIC	5 (6)
Postinduction G stageb
GX	4 (5)
G0	18 (23)
G1	2 (3)
G2	36 (45)
G3	11 (14)
G4	9 (11)
Pathological T statusb
T0	18 (23)
Tis	1 (1)
T1	28 (35)
T2	7 (9)
T3	25 (31)
T4	1 (1)
Pathological N statusb
N0	48 ( 60)
N1	18 (23)
N2	10 (13)
N3	4 (5)

^aPreinduction denominator: 114 patients enrolled.

^bPostinduction denominator: 80 patients undergoing an operation.

Dysphagia scores were recorded for all patients at their initial consultation (Table 5). Fifty-four percent (61/114) had dysphagia of any grade, and 46% (53/114) did not have dysphagia. Among the patients who had dysphagia, 66% (40/61) had grade 1, 25% (15/61) had grade 2, and 10% (6/61) had grade 3 or 4 dysphagia. Among these patients, 89% (54/61) had T3-4 disease by EUS. In contrast, only 53% of patients without dysphagia (28/53) had T3-4 disease by EUS (p<0.001). In addition, for the 9% (10/114) of patients with cT1-2 with cN1 disease, dysphagia was present in 30% (3/10) which was not predictive of N+ disease.

Table 5.

Comparison of Dysphagia Score to Endoscopic Ultrasound (EUS) T Stage

Dysphagia Score	No. (%)
0	53 (46)
1	40 (35)
2	15 (13)
3	5 (4)
4	1 (1)

	Dysphagia, No. (%)
EUS T Stage	No	Yes	Total
3 or 4	28 (53)	54 (89)	82
2 or less	25 (47)	7 (11)	32
Total	53	61	114

Dysphagia score: 0, able to eat normal diet; 1, able to eat some solid food; 2, able to eat semisolids only; 3, able to swallow liquids only; 4, unable to swallow.

Comment

The initial staging evaluation of patients with newly diagnosed esophageal cancer is extensive. After a diagnostic esophagogastroduodenoscopy, the evaluation includes a computed tomography scan of the chest, abdomen, and pelvis, positron emission tomography scan, and EUS. In the absence of metastatic disease, the EUS is considered critical to determine which patients should be treated with neoadjuvant therapy followed by resection and which should proceed directly to resection. Although the decision of whether to administer neoadjuvant therapy can vary by patient, most patients with T3 or greater tumors receive neoadjuvant therapy before resection [6-8]. The distinction between T3 lesions and <T3 lesions is typically made by means of EUS, a test that is commonly performed separately, following the initial diagnostic esophagogastroduodenoscopy, and that can frequently lead to a delay in the initiation of treatment [2]. Given the time-consuming nature of esophageal cancer evaluation, eliminating a potentially unnecessary diagnostic test may result in more efficient initiation of therapy.

Dysphagia is a common presenting symptom that is readily identified during initial history and physical examination. In this prospective series, 54% of patients with esophageal cancer (61/114) presented with dysphagia of any grade; therefore, decisions based on this symptom would apply to approximately half of patients with esophageal cancer. Previously, our institution showed, in a retrospective analysis, that patients with dysphagia often have a higher T stage [4]. This observation led to the question of whether EUS adds to the management of patients with dysphagia or whether it is an unnecessary test for this group of patients.

In the present study, 89% of patients with dysphagia of any grade (54/61) had cT3-4 disease by EUS. Although most patients had grade 1 dysphagia (66%), any dysphagia was predictive of cT3-4 disease. In contrast, lack of dysphagia did not correlate with depth of invasion, as assessed by EUS. Among patients without dysphagia, 53% (28/53) had cT3-4 disease and 47% (25/53) had cT1-2 disease.

Although forgoing a staging EUS in patients with any dysphagia may seem to be unnecessarily parsimonious, the predictive value of dysphagia is 89% for cT3-4 disease. Rice and colleagues reported that, for >T2 lesions, EUS had an accuracy of 87% and a positive predictive value of 86% for pathological staging[8]. The positive predictive value for T2 lesions was significantly worse, at 23%. O'Farrell and colleagues reported a lower accuracy for EUS assessment of T stage of 71% [9]. Ideally, the accuracy of T stage identification would be better than 89%, but assessment by dysphagia alone is at least as accurate as EUS in clinical staging. In addition, occasionally patients with dysphagia may need an EUS to assess nodal disease or local invasion if CT and or PET/CT results are equivocal.

There are several limitations to this study. The majority (96%) of patients had adenocarcinoma; only 4% had squamous cell carcinoma. Although adenocarcinoma is the histologic profile increasing most in the United States, the proportion of adenocarcinoma in our study may not represent all practices, and our findings may not be as relevant for squamous cell carcinoma. Despite the highly predictive nature of dysphagia, 11% of patients with dysphagia had disease that was staged at T2 or less; therefore, induction therapy may constitute overtreatment for some patients. Lastly, lack of dysphagia was not predictive of early-stage disease; therefore, staging EUS should still be used in patients without dysphagia.

In conclusion, because the presence of dysphagia in patients with esophageal cancer is highly predictive of T3-4 disease, consideration should be given to omitting EUS and proceeding directly to neoadjuvant therapy for these patients. Approximately half of patients currently undergoing a staging EUS could potentially forgo EUS before neoadjuvant therapy. On the other hand, for patients without dysphagia, the likelihood of having T1-2 versus T3-4 disease is approximately 50%; therefore, EUS should remain a routine part of the evaluation of these patients.