Recurrence after Neoadjuvant Chemoradiation and Surgery for Esophageal Cancer: Does the Pattern of Recurrence Differ for Complete Responders and Those with Partial or No response?

Abstract

Objective

We hypothesized that most relapses in esophageal patients undergoing neoadjuvant chemoradiation therapy would occur outside of the surgical and radiation fields.

Methods

Recurrence patterns, time to recurrence, and median survival were examined in 267 patients who underwent esophagectomy after neoadjuvant chemoradiation therapy at Johns Hopkins over 19 years.

Results

Of 267 patients, 82 (30.7%) were complete responders (CR) to neoadjuvant therapy, with 108 (40.4%) and 77 (28.8%) being partial responders (PR) and non-responders (NR), respectively. Recurrence developed in 84 patients, (CR 18 /82 (21.4%), PR 39/108 (36.1%), NR 27/77 (35.1%) ,p=0.055, respectively.) Most patients recurred at distant sites (65/84;77.4) regardless of pathological response and subsequent survival was brief (median 8.37 months). Median disease-free survival was short (10 months) and did not differ based on recurrence site for partial and non-responding patients, but was longer for CR patients with distant recurrence whose median disease-free survival was 27.3 months (p=0.008).By multivariate analysis, no other factor except for pathological response to neoadjuvant therapy was associated with disease recurrence or death. PR and NR patients were 1.97 and 2.23 times more likely to recur than CR patients (p=0.024 and p=0.012, respectively).

Conclusions

Most esophageal cancer recurrences after neoadjuvant therapy and surgery are distant, and survival time after recurrence is short regardless of pathological response. Fewer patients achieving CR recur, and distant recurrences in these patients manifest later than in patients with PR and NR. Only pathological response is significantly associated with disease recurrence suggesting that tumor biology and chemosensitivity are critical in long-term patient outcome.

Introduction

Most surgical candidates with esophageal cancer present with regional nodal metastases, or stage III disease.¹ Historically, the primary curative treatment was a total or partial esophagectomy and lymph node dissection with resultant five-year survival rates ranging between 15% and 20%.¹ After encouraging preliminary data in the late 1980s, and our own investigations, neoadjuvant chemoradiotherapy in combination with surgery became the mainstay of treatment at our institution over the last 19 years. This trimodality approach is now widely accepted clinical practice.^{2, 3}

Despite this widespread use especially in North America, short and long-term patient outcome data after neoadjuvant therapy plus surgery have been conflicting.^{4, 5} Importantly however, the 25-30% of patients with a pathologic complete response (CR) after this therapy have significantly improved 5-year survival rates compared to other treatment modalities for advanced esophageal cancer.^4-7 However, it is also clear that about 30% of patients with a pathologic CR, still recur and die from metastatic esophageal cancer⁸. In these pathologic CR patients, in particular, little is known about the impact of neoadjuvant chemoradiation on the timing and patterns of recurrent esophageal cancer or on the overall survival of these patients with recurrent disease.

We hypothesized that chemoradiation plus surgery would lead to improved local tumor control compared to surgery alone and that most relapses would occur outside the surgical and radiation fields. We also hypothesized that recurrence would predominate in patients with chemoresistant tumors that were not impacted by neoadjuvant chemoradiation and surgery, and thus would occur relatively quickly. We, therefore, examined if recurrence patterns, time to recurrence, and overall median survival differed with pathologic response to neoadjuvant chemoradiotherapy.

Methods

Data Source

This is a retrospective analysis using patient data collected from the Johns Hopkins Hospital Multidisciplinary Esophageal Cancer Database which contains demographic, clinical and patient outcome on esophageal cancer patients evaluated at Johns Hopkins Hospital between January 1, 1989 and December 31, 2007. Data in the database were obtained from patient hospital charts, as well as electronic and paper files. Since 2000, periodic data audits have been performed by archival and on-line chart reviews for quality assurance. These reviews have consistently verified over 90% accuracy of the database with source materials.

Patient Outcome Data

Follow-up data and sites of tumor recurrence were determined by a systematic follow-up of any patients with incomplete data attained by letters, emails, faxes and phone calls to their referring physicians. In almost all cases, recurrence was documented radiographically, but, in a few cases, there was also pathological assessment. All sites of first recurrence were recorded. These data were obtained from clinic charts with confirmation by reviewing associated imaging and pathology reports. This study was approved by the Johns Hopkins Institutional Review Board, who exempted the need for patient consent, and abides by Health Insurance Portability and Accountability Act compliance standards.

Patient Population

Patient selection criteria included a biopsy proven esophageal carcinoma and a documented assessment by a thoracic surgeon, a medical oncologist, a gastroenterologist, and a radiation oncologist at the Johns Hopkins Hospital. We identified 267 patients who underwent pre-treatment clinical staging, neoadjuvant chemotherapy and/or radiotherapy, and had an esophagectomy with curative intent.

Race was self-reported as Caucasian, African American or other. For ease of analysis, patients of other race were documented as minorities and combined with the African American patients. Clinical staging was determined at the time of initial evaluation and was available on all but 2 patients who both were complete responders. All 267 patients had a recorded pathologic stage derived from the pathology record. Smoking history was available on all but 5 patients. Patients were identified as ever or never smokers based on self-reporting. Ever smokers were defined as those who had smoked 100 tobacco cigarettes or more during their lifetime, while never smokers were those who had smoked fewer than 100 cigarettes. Pack year index was determined by multiplying the average number of cigarette packs smoked per day by the number of years smoked. The American Society of Anesthesiologists (ASA) 5-grade classification system was used as an index of preoperative comorbidity.⁹ Esophagectomies were performed as partial esophagectomies using the transhiatal (81%), the 3-incision (15%), the Ivor-Lewis (2%), or the thoracoabdominal approach (2%) (Table 1). During the length of the study, these four approaches were all used with concomitant mediastinal lymphadenectomy. The median number of lymph nodes harvested was 15.

Table 1.

Characteristics of the Study Population by Pathological Response N=267

Table 1	Complete Response 82 (30.7%)		Partial Response 108 (40.4%)		No Response 77 (28.8%)		p value
	#	%	#	%	#	%
Age at Diagnosis, (yrs), median & IQR	61	51 - 67	61	51 - 67	60	52 - 67	0.773
Sex							0.002
Male	66	80.5	104	96.3	69	89.6
Female	16	19.5	4	3.7	8	10.4
Race							0.404
White	76	92.6	102	94.4	68	88.3
Black	3	3.7	4	3.8	7	9.1
Other	3	3.7	2	1.8	2	2.6
Smoked Cigarettes*							0.952
Never	15	19.0	22	20.4	14	18.7
Ever	64	81.0	86	79.6	61	81.3
Pack-Years Smoked, (yrs) *  (median & IQR)	32.9	23.7 - 53.6	35	15 - 50	40	25 - 60	0.284
ASA Classification, median	3		3		3		0.565
Histology							<0.001
Adenocarcinoma	54	65.9	96	88.9	58	75.3
Squamous Cell	24	29.3	12	11.1	19	24.7
Carcinoma, NOS	4	4.8	0	0.0	0	0.0
Clinical Stage#							<0.001
IIA	29	36.3	11	10.2	33	42.9
IIB	9	11.2	15	13.9	15	19.4
III	35	43.8	61	56.5	23	29.9
IVa	7	8.7	21	19.4	6	7.8
Pathologic Stage							<0.001
No Evidence of Disease	82	100	0	0.0	0	0.0
I	0	0.0	30	27.8	0	0.0
IIA	0	0.0	60	55.6	27	35.0
IIB	0	0.0	13	12.0	9	11.7
III	0	0.0	5	4.6	35	45.5
IVa	0	0.0	0	0.0	2	2.6
IVb	0	0.0	0	0.0	4	5.2
Neoadjuvant Chemotherapy							0.149
5-FU and Platinum Only	62	75.6	70	64.8	46	59.7
Non-5-FU/Platinum	8	9.8	21	19.5	19	24.7
Course Unknown	12	14.6	17	15.7	12	15.6

Neoadjuvant Chemoradiation and Pathological Response

In order to compare and control for multimodal pre-operative and post-operative therapy, we examined neoadjuvant chemo- and radiotherapy types, doses as well as adjuvant chemotherapy. Neoadjuvant chemotherapy agents were almost always given as platinum-based doublets and are stratified as either with or without 5-FU. Neoadjuvant radiotherapy dose was recorded as either <4,400 cGy or >= 4,400 cGy. Post-operative adjuvant chemotherapy was recorded as either administered or not.

Pathological response to neoadjuvant chemoradiation was based on our previously described classification system and determined at the time of pathological examination of the surgical specimen¹⁰. Briefly, patients were identified as having a pathologic CR if no microscopic evidence of tumor was found upon examination of both the resected esophageal specimen and nodal tissues. Partial response (PR) described patients with persistence of microscopic esophageal carcinoma in the resected surgical specimen but their overall AJCC¹¹ stage was lower as compared to pre-operative clinical staging. Non-response (NR) patients had either no change in AJCC stage between clinical and pathological staging, or progression of disease in spite of neoadjuvant therapy. Of note, at our institution, we treat patients with Stage IVA disease (local lymph node involvement) with multimodality therapy. Preoperative clinical staging was performed by compiling results from CT scans, esophageal endoscopic ultrasound (EUS), barium esophagrams, and in some patients, exploratory laparoscopy¹⁰.

Recurrences

Time between diagnosis and surgery, length of stay after surgery, recurrence-free survival (time from surgery to disease recurrence), and overall survival were calculated and reported. All patients in the dataset were included in consideration of disease recurrence (or disease-progression for the patients with no response to neoadjuvant chemoradiation) except 6 patients who were found at time of surgery to have metastatic stage IVa (2 patients with celiac lymph nodes) or IVb disease (2 patients with isolated liver metastases, 1 patient with an omental metastasis, and 1 patient with supraclavicular disease). Note that these 6 patients still were aggressively resected and local therapy applied to their metastatic site, and so they are included in the larger cohort of 267 patients. All sites of tumor recurrence were recorded and categorized as the abdomen (excluding the liver), the liver, the brain, bone, the chest (excluding the esophagus), the esophagus, and other.

Statistical Analysis

Comparison of continuous, categorical and dichotomous variables was performed using the Student’s t test and chi-squared test for homogeneity. Time to recurrence-free survival and all cause mortality was modeled using the Kaplan-Meier method, and the association of factors with time to recurrence and death was analyzed using the Cox Proportional Hazards Model. The model was adjusted by risk factors for recurrence and mortality that included patient sex, histology, and clinical stage. Crude hazard ratios (HR) and corresponding 95% confidence intervals are reported. All HR are presented as independent variables, and then further adjusted for specified variables and presented as adjusted hazard ratios.

Overall survival was defined as time from date of surgical therapy to death or last follow-up. Similarly, disease-free survival was defined as time from date of surgical therapy to first recurrence or last follow-up.

Differences between Kaplan-Meier survival curves were estimated using the log-rank test. All comparisons with two-sided p-values ≤0.05 were considered statistically significant. Statistical analysis was performed using the software package STATA 10.0 (StataCorp, College Station, TX).

Results

Table 1 shows the clinicopathological characteristics of all 267 patients. Median age was 60 years IQR (51 - 67) with 239 (89.5%) being male. The racial distribution included 246 (92.2%) Caucasians, 14 (5.2%) African Americans, and 7 (2.6%) of other race. Most patients, 211 (80.5%), were ever smokers (87 current smokers; 124 former smokers) with a median pack-years smoked of 35. Adenocarcinoma was the reported histology for 208 (77.9%) patients, squamous cell histology for 55 (20.6%) patients, and 4 (1.5%) patients with carcinoma, not otherwise specified. Median time between diagnosis and surgery was not different between response groups nor was there a difference in median hospital length of stay. In patients with known chemotherapy regimens and radiation doses, the majority received 5-FU and platinum neoadjuvant chemotherapy (178/226, 78.8%), along with at least 4,400 cGy (151/166, 91%) of neoadjuvant radiation. Only 56/267 (21.0%) of patients received adjuvant chemotherapy.

Of 267 patients, 82 (30.7%) had a CR to neoadjuvant therapy, 108 (40.4%) patients had a PR, and 77 (28.8%) had NR. On univariate analysis, PR, CR and NR cohorts differed significantly by sex, tumor histology, and initial clinical stage (Table 1). A total of 84 patients (31.5%) developed recurrence of their esophageal malignancy. There was a strong trend towards a statistically significant difference in frequency of esophageal recurrence between response groups (CR 18/82, (22.%), PR 39/108, (36.1%), and NR 27/77, (35.5%)) (p=0.055) (Table 2). The distributions of patterns of first tumor recurrence for the individual organ sites between the three different pathological response groups are also presented in Table 2. Although there seems to be proportionately fewer bone/liver metastases for the CR patients, there is no significant difference in site of recurrences between response groups. When recurrence sites were grouped as either within the radiation field (including esophagus and mediastinum) versus outside the field (all other sites of recurrence), again there was no significant difference in patterns of recurrence according to the pathological response group (Table 3).

Table 2.

First Tumor Recurrence Site According to Pathological Response Status to Neoadjuvant Therapy (n=84)

Tumor Site	Proportion of Patients Recurring from Total Cohort
	Complete Response (n=82)		Partial Response (n=108)		No Response (n=77)
	#	%	#	%	#	%	p value
Overall Recurrence	18	22.0	39	36.1	27	35.1	0.055
Esophagus and Mediastinum	5	27.2	8	20.5	6	22.2
Lung and Pleura	4	22.2	7	17.9	4	14.8
Abdomen	5	27.7	9	23.0	4	14.8
Liver	1	5.5	6	15.3	6	22.2
Brain	1	5.5	2	5.1	2	7.4
Bone	2	11.1	7	17.9	5	18.5

Table 3.

Proportion of First Recurrences Within and Outside the Radiation Field According to Pathological Response (N=84)

Radiation Field	Proportion Recurred
	Complete Response N=18		Partial Response N=39		No Response N=27
	#	%	#	%	#	%	p value
Within Radiation Field	5	27.8	8	20.5	6	22.2	0.829
Outside Radiation Field	13	72.2	31	79.4	21	77.8

Median disease-free survival differed significantly when stratified by response group and whether or not the first site of recurrence was locoregional or distant (Table 4). All recurrences occurred within a median time of 10 months postoperatively except for distant recurrences in patients who had a CR to neoadjuvant chemoradiation. In these patients the median disease-free survival was 27.3 months (p=0.008). Thus, these data suggest that neoadjuvant chemoradiation delayed recurrence significantly and as well fewer patients failed (p=0.055). Fig. 1 showing freedom from recurrence again underscores the fact that patients with CR have a significantly longer disease-free interval than patients with PR and NR (p=0.01).

Table 4.

Median Disease-Free Survival of Patients with Recurrence (N=84)

	Complete Response		Partial Response		No Response		P value
	Median	IQR	Median	IQR	Median	IQR
Within Radiation Field	5.8	4.7 - 9.3	7.1	3.5 - 18	8.5	7.5 - 62.3	0.604
Outside Radiation Field	27.3	12.3 - 31.1	5.7	3.9 - 15.8	9.1	6.9 - 14.5	0.008

Figure 1.

Kaplan-Meier Estimates of Recurrence-Free Survival of Esophageal Cancer Patients (n=261) at the Johns Hopkins Hospital, according to Pathological Response at Time of Surgical Resection.

This impact of chemoradiation in preventing recurrence was of sufficient magnitude also to impact survival. The median overall survival was greatest for the CR cohort (79.3 months), which was significantly longer than that of the PR (30.6 months) and NR cohorts (18.6 months), respectively (P<0.001, Table 1). Kaplan-Meier overall survival estimates of all three pathological response groups are depicted in Fig. 2 showing a statistically significant difference in overall survival (all-cause mortality) at 5 years. Interestingly, early mortality (<10 months) did not differ by pathological response group with only patients surviving longer than 1 year showing significant differences in mortality by pathological response to neoadjuvant therapy (Fig. 2). The overall 5-year survival for CR and PR patients who did not recur is statistically longer than for those patients who recurred with cancer (Fig. 3; Fig. 4). NR patients who did not progress with further disease had survival outcomes as poor as those who showed clinical evidence of disease progression (Fig. 5). This is probably because if tumors are aggressive enough to be recalcitrant to neoadjuvant chemoradiation, little additional tumor burden is necessary to cause eventual death.

Figure 2.

Kaplan-Meier Estimates of Overall Survival of Esophageal Cancer Patients (n=267) at the Johns Hopkins Hospital, according to Pathological Response at Time of Surgical Resection. The Kaplan-Meier estimates for overall survival indicate that as the extent of pathological response to neoadjuvant therapy increases, there is a significant increase in the overall survival of the patient. It is interesting to note, however, that in our observed cohort, the degree of pathological response to neoadjuvant therapy seemed not to affect patients with early mortality (<10 months) but rather those who survived more than one year.

Figure 3.

Kaplan-Meier Estimates of Overall Survival of Esophageal Cancer Patients who had a complete pathological response to neoadjuvant therapy (n=82) at the Johns Hopkins Hospital, according to whether or not there was cancer recurrence following surgical resection.

Figure 4.

Kaplan-Meier Estimates of Overall Survival of Esophageal Cancer Patients who had a partial pathological response to neoadjuvant therapy (n=108) at the Johns Hopkins Hospital, according to whether or not there was cancer recurrence following surgical resection.

Figure 5.

Kaplan-Meier Estimates of Overall Survival of Esophageal Cancer Patients who had no pathological response to neoadjuvant therapy (n=71) at the Johns Hopkins Hospital, according to whether or not there was cancer recurrence following surgical resection.

After adjusting for sex, histology and clinical stage in a multivariate model, no other demographic or clinicopathologic factor except for pathological response to neoadjuvant therapy at time of surgical resection was associated with disease-free survival. PR and NR patients were 1.97 and 2.23 times more likely to recur (p=0.024 and p=0.012, respectively)(Table 5). Once a patient recurred, all deaths occurred within a year regardless of initial response groups (Table 1) and all deaths were cancer-related (data not shown). Finally, the median overall survival of patients who recurred vs. patients who did not recur was 21.2 months vs. 65.2 months, respectively; p<0.01.

Table 5.

Crude and Adjusted Hazard Ratios for the Association between Pathological Response and Recurrence

	Crude Hazard Ratio	95% CI	p value	Adjusted Hazard Ratio*	95% CI	p value
Clinical Response
Complete Response	1.00	referent		1.00	referent
Partial Response	2.13	1.22 - 3.74	0.008	1.97	1.10 - 3.57	0.024
No Response	2.26	1.24 - 4.13	0.008	2.23	1.20 - 4.17	0.012
Age at Diagnosis, (yrs)	0.98	0.96 - 1.00	0.164
Sex
Male	1.00	referent		1.00	referent
Female	0.74	0.34 - 1.60	0.440	0.91	0.38 - 2.21	0.840
Race
White	1.00	referent
Black	1.58	0.58 - 4.34	0.369
Histology
Adenocarcioma	1.00	referent		1.00	referent
Squamous Cell	0.85	0.45 - 1.43	0.458	0.95	0.50 - 1.83	0.887
ASA Classification
2	1.00	referent
3	0.97	0.50 - 1.88	0.925
4-5	0.8	0.22 - 2.91	0.734
Clinical Stage
IIA	1.00	referent		1.00	referent
IIB	1.42	0.74 - 2.72	0.290	1.21	0.63 - 2.35	0.565
III	1.21	0.72 - 2.04	0.469	1.19	0.68 - 2.07	0.537
IVa	1.18	0.53 - 2.65	0.681	1.13	0.47 - 2.68	0.787
Neoadj Chemotherapy
5-FU Based	1.00	referent
Non-5-FU Based	0.90	0.53 - 1.51	0.684
Neoadj Radiotherapy
≥4,400 cGy	1.00	referent
<4,400 cGy	0.47	0.15 - 1.52	0.209
Dose Unknown	0.85	0.53 - 1.36	0.497
Smoked Cigarettes
Never	1.00	referent
Ever	1.09	0.63 - 1.88	0.755
Pack-Years Smoked	0.99	0.98 - 1.01	0.250

^*Adjusted for sex, histology, and clinical stage

Discussion

We report on our institution’s experience with esophageal cancer, examining survival outcomes and patterns of disease recurrence after neoadjuvant chemo-radiotherapy. For all three response groups, most recurrences were distant, suggesting the important role of intensified local therapies to eradicate local disease, and prevent locoregional recurrence. By multivariate analysis, only pathological response to neoadjuvant chemotherapy is a significant factor in disease recurrence. The implication of this finding is that, given eminent resectability of the primary tumor, the esophageal tumor biology with its intrinsic sensitivities to the correct chemoradiation regimens is the most critical factor in determining patient outcome. Complete responders, even those who eventually recur with distant disease, experienced the longest median recurrence-free survival time, and moreover, patients with a complete response to neoadjuvant therapy had significantly improved overall survival compared to the partial and non-responders. The frequency of recurrence tended to be greater for non-responders, and lowest for complete responders.

In CR patients, tumors exhibited such a high degree of chemoradiation sensitivity that not only the primary tumor, but also distant micrometastases were eliminated by the neoadjuvant therapy. It is possible that in the PR patients with variable tumor sensitivity, the persistence of distant micrometastases caused distant tumor recurrence. In an aggressive cancer, such as esophageal cancer, any remaining distant metastasis after neoadjuvant therapy ensured that PR patients recurred outside of the treated field with the same virulence as those NR patients not affected by neoadjuvant therapy (Table 4).

These findings reinforce just how critical a CR can be to an esophageal cancer patient receiving neoadjuvant therapy and surgery. The 5 CR patients whose recurrence was local received no discernable benefit from chemoradiation as the median disease-free interval of these 5 patients was not different from that of the patients who had a PR/NR response. This is in direct contrast to the 13 CR patients who recurred distantly after neoadjuvant therapy and surgery since these patients had late relapse with a median recurrence time of over 2 years. Presumably, the micrometastatic disease present in these 13 CR patients had a molecular profile with some degree of chemosensitivity to the preoperative chemotherapeutic regimen given, which presumably delayed the eventual re-emergence of distant macrometastases.

The demographic and histopathological composition of our cohort is representative of other large medical centers with high volume esophageal practices in the USA over this time period^{6, 7, 12}Our percentages of CR, PR and NR esophageal cancer patients who received neoadjuvant therapy is also very similar to that of other groups^{6, 7} despite small differences in definitions used for codification of groups. The similarity of our results to others validates the universal nature of our findings.

Although our data showed that locoregional recurrence rates did not differ according to response groups (Table 3), we do not think that these data can be extrapolated to support the assertion that surgery alone may be effective at controlling local disease. Our 23% local recurrence rate for patients receiving neoadjuvant chemoradiation closely resembles the 19% local failure rate of the neoadjuvant chemoradiation arm of the randomized clinical trial of Urba et al. conducted at the University of Michigan¹³. Moreover, in both the Michigan trial and the large, randomized U.S. G-I Intergroup trial, the locoregional disease recurrence rates for the surgery alone control arms were 42% and 58%, respectively^{13, 14}. Importantly, our paper therefore further corroborates the point that surgery alone, nor chemoradiation alone, is as effective as trimodality therapy in achieving local disease control.

In multivariable regression analysis examining the likelihood of tumor recurrence, patients with a partial response or non-response to neoadjuvant therapy were 1.97 and 2.23 times more likely to develop tumor recurrence compared to CR patients. These data are consistent with varying degrees of tumor resistance to cytotoxic agents and radiotherapy, and correlate with shortened survival and death. Thus pathological response status was a reasonable surrogate for risk of recurrence and may be useful along with molecular characteristics in the future for selecting patients for additional or alternative forms of therapy following surgery.

Limitations of this study include its retrospective nature, the confines of a single institutional study, and limitations of the database. Hence, specific data points such as cancer-specific death were not able to be obtained on all patients. This limits the ability to determine with certainty that death during follow-up is entirely due to recurrence of tumor. The relatively small size of this study does not allow for rigorous stratification, such as examining differences in outcomes between chemotherapeutic agents and different doses of radiotherapy. In larger databases, however, such as the Nationwide Inpatient Sample and the Surveillance, Epidemiology, and End Results databases, these variables are either characteristically missing or erroneous. Ideally, in order to study outcomes such as those we report, a multi-institutional comprehensive patient-oriented database would be needed. Since this study spans 19 years and multiple examining pathologists, we reported pathological findings based on a simple standard of CR, PR and NR rather than the more detailed descriptor of percentage of viable cells in the specimen. A final limitation of the study is that because of its retrospective nature, the time to recurrence (and location of recurrences) could have been impacted by the fastidiousness of monitoring.

The findings of this study, and others¹⁵, suggest that not all esophageal carcinomas are equal, and that rather than more aggressive chemotherapy regimens, or more intensive radiation dosing schedules, efforts must be made to target different tumors with chemotherapeutic agents to which they are sensitive. This new era of “personalizing” or customizing chemotherapeutic and radiation regimens to specific molecular targets in the primary tumor is now upon us, and because of the success of the neoadjuvant therapy approach in esophageal cancer, this is an appropriate cancer model to test these innovations. Importantly, this study emphasizes the critical role of neoadjuvant chemotherapy in “eliminating” distant disease outside of the local surgery and radiation fields, and offers yet more evidence that ultimately the control of distant disease is critical in preventing esophageal cancer recurrence. Patients who experience complete response to neoadjuvant therapy have the best outcomes in terms of lowest incidence of recurrence, longest disease-free survival, and longest overall survival.