Induction Chemotherapy Plus Neoadjuvant Chemoradiation for Esophageal and Gastroesophageal Junction Adenocarcinoma

Abstract

Background:

Neoadjuvant chemotherapy with concurrent radiotherapy (nCRT) is an accepted treatment regimen for patients with potentially curable esophageal and gastroesophageal junction (GEJ) adenocarcinoma. The purpose of this study was to evaluate if induction chemotherapy (IC) before nCRT is associated with improved pathologic complete response (pCR) and overall survival (OS) when compared with patients who received nCRT alone for esophageal and GEJ adenocarcinoma.

Methods:

Using the National Cancer Database (NCDB), patients who received nCRT and curative intent esophagectomy for esophageal or GEJ adenocarcinoma from 2006–2015 were included. Chemotherapy and radiation therapy start dates were used to define cohorts who received IC before nCRT (IC + nCRT) versus those who only received concurrent nCRT before surgery. Propensity weighting was conducted to balance patient, disease, and facility covariates between groups.

Results:

12,460 patients met inclusion criteria. 11,880 (95%) patients received nCRT and 580 (5%) patients received IC + nCRT. Following propensity weighting, OS was significantly improved among patients who received IC + nCRT versus nCRT (HR 0.82; 95% CI 0.74–0.92; p<0.001) with median OS for the IC + nCRT cohort of 3.38 years versus 2.45 years for nCRT. For patients diagnosed from 2013–2015, IC + nCRT was also associated with higher odds of pCR compared to nCRT (OR 1.59; 95% CI 1.14–2.21; p=0.007).

Conclusion:

IC + nCRT was associated with a significant OS benefit as well as higher pCR rate in the more modern patient cohort. These results merit consideration of a sufficiently powered prospective multi-institutional trial to further evaluate these observed differences.

INTRODUCTION

Esophageal adenocarcinoma is a particularly lethal malignancy, and its incidence continues to rise (1). A number of randomized controlled trials over the past two decades have demonstrated superior survival benefits of neoadjuvant and perioperative therapy over surgery alone for resectable esophageal and gastric cancers (2–5). While surgery remains essential to curative treatment, multimodal therapy with either concurrent neoadjuvant chemoradiation (nCRT) or perioperative chemotherapy has become an integral part of the treatment regimen and is now considered the standard of care (6). Despite these advances, survival of esophageal cancer remains poor, and improvements in treatment approaches are desperately needed.

Since patients tend to die of systemic rather than local disease, one strategy to improve systemic control and long-term outcomes is to utilize induction chemotherapy (IC) plus nCRT (IC + nCRT). Several retrospective studies have suggested that the addition of IC to nCRT may improve pathologic complete response (pCR), overall survival (OS), and disease-free survival (DFS), especially in patients with more advanced esophageal cancer (7–9). This strategy has also been employed in other gastrointestinal cancers with promising results (10). For example, a recent systematic review showed that total neoadjuvant therapy with a combination of IC and nCRT for rectal cancer increased pCR by 39% compared with conventional nCRT (11). Additionally, the use of IC may aid in personalizing care, as the use of PET/CT following IC can be both prognostic and help guide chemotherapy drug choice during nCRT to improve pCR rates and OS (12–14).

In 2013, Ajani et al. reported the results of a phase II randomized controlled trial evaluating the effect of IC in patients that underwent trimodal therapy and concluded that there was no statistically significant benefit in OS or pCR with the addition of IC over nCRT alone (15). However, this study may have been underpowered, as a secondary analysis performed suggested that IC + nCRT may confer an OS benefit in the subset of patients with well to moderately differentiated esophageal cancer (16).

The objective of this study was to overcome some of the limitations from smaller studies by evaluating real world outcomes from a large national database to compare the treatment effects of IC + nCRT versus nCRT alone for resectable esophageal and gastroesophageal junction adenocarcinoma. Our study uses sequential timing of chemotherapy before radiation as a surrogate for the use of IC prior to nCRT.

METHODS

Study Design

This is a retrospective cohort study using the NCDB esophageal and gastric participant user files. The NCDB is a joint program of the Commission on Cancer of the American College of Surgeons and the American Cancer Society and consists of patients who received treatment at a Commission of Cancer accredited facility (17). The NCDB captures approximately 70% of all new cancer diagnoses in the United States (18).

Adult patients with esophageal or GEJ adenocarcinoma diagnosed between 2006 and 2015 who received neoadjuvant chemotherapy and radiation therapy followed by esophagectomy were included. Exclusion criteria included unknown neoadjuvant therapy start dates, metastatic or T4 disease, patients who underwent a palliative esophagectomy, and those who underwent esophagectomy at a facility other than the NCDB reporting facility of record. Additionally, patients were excluded if neoadjuvant radiation therapy started 6 weeks prior to or 6 months after neoadjuvant chemotherapy start date.

Patients were categorized into two cohorts based on the timing of receipt of chemotherapy and radiation therapy from their date of diagnosis. Patients who received radiation treatment starting greater than 6 weeks after their chemotherapy start date were classified as receiving IC + nCRT. Patients who received radiation treatment starting within 6 weeks of their chemotherapy start date were classified as receiving nCRT.

The outcomes of OS and pCR rate were compared between patients who received IC + nCRT versus nCRT. OS was calculated from the date of surgery to the date of last follow up or death. pCR was defined as pathologic staging of pT0 and pN0 based on the traditional AJCC staging system. Patients with unknown or missing pathologic staging were excluded.

Statistical Analysis

Demographic and tumor characteristics were compared between patients who received IC + nCRT versus nCRT. Continuous variables were expressed by means and standard deviations, and categorical variables were expressed as absolute numbers and percentages. Differences between the two treatment groups were assessed using chi-square test and Student t-test for categorical and continuous variables, respectively.

A weighted propensity score analysis was performed to balance covariates between the two treatment groups (19, 20). Propensity scores were calculated using a multivariable logistic regression model in which treatment (IC + nCRT versus nCRT) was regressed on demographic and tumor characteristics. The propensity score was calculated as the predicted probability of receiving IC + nCRT. Stabilized inverse probability of treatment weighting was then conducted using the propensity score (21), and covariate balance was checked (Table 1).

Table 1:

Patient and disease characteristics of the study population (2006–2015) (N=12,460).

	N	%
Sex
Male	10,949	87.87
Female	1,511	12.13
Age (mean, SD)	62.1	9.4
Race
White	11,979	96.14
Black	238	1.91
Other	243	1.95
Charlson/Deyo Score
0	8,704	69.86
1	2,916	23.40
2–3	840	6.74
Tumor Location
Esophagus	8,264	66.32
GE Junction	3,453	27.71
Unknown	743	5.96
Clinical T Stage
T1	611	4.90
T2	2,211	17.74
T3	8,241	66.14
Unknown	1,397	11.21
Clinical N Stage
N0	4,284	34.38
N1	5,942	47.69
N2–N3	1,286	10.32
Unknown	948	7.61
Histologic Grade
Well-Moderately Differentiated	4,779	38.35
Poorly Differentiated	5,935	47.63
Unknown	1,746	14.01
Facility Type
Community	217	1.74
Comprehensive	3,069	24.63
Academic/Research	7,457	59.85
Integrated	1,717	13.78

The Kaplan-Meier method was used to estimate overall survival between treatment groups before and after propensity score weighting. A log-rank test was used to test the differences between the Kaplan-Meier distributions for the unweighted sample and a univariate Cox proportional-hazard model was used for the weighted sample. A propensity score weighted Cox-proportional hazards model was used to estimate the adjusted hazard ratios (HRs) for OS. The following variables were included in the Cox-proportional hazards model: treatment group (IC + nCRT versus nCRT), sex, age at diagnosis, race, tumor location, clinical T stage, clinical N stage, histologic tumor grade, facility type, and year of diagnosis (2006–2012 versus 2013–2015). Year of diagnosis was categorized into two time periods, early (2006–2012) and late (2013–2015), to account for temporal trends in chemotherapy regimens before and after publication of the CROSS trial (3). The two-way interaction between treatment cohort and time period of diagnosis was included to test for a different impact of treatment group on OS in these two time periods. Additionally, the two-way interaction between treatment group and histologic tumor grade was included to test for a different impact of treatment based on tumor grade. Both interactions were excluded from the final model, as they were not significant.

For the outcome of pCR, after excluding patients with unknown or missing data, a weighted propensity score analysis was again performed to balance covariates between the two treatment groups. A propensity score-weighted logistic regression model was used to examine the adjusted association between treatment group and pCR, controlling for the same variables as described above for the survival analysis. Two-way interactions between treatment group and time period and tumor grade were again evaluated. The two-way interaction between treatment group and time period was statistically significant and included in the final model.

For all Cox-proportional hazards models, the proportional hazards assumption was checked for each variable graphically using log-log plots. A 2-sided significance level of 0.05 was used for all statistical tests. All analyses were performed using STATA version 15.1 (STATA Corp, College Station, TX).

RESULTS

Study Population

From 2006–2015, 33,332 patients who underwent esophagectomy for esophageal or GEJ adenocarcinoma were identified in the NCDB. Receipt of neoadjuvant therapy was unknown or missing in 5,619 patients. There were 10,723 patients (32.2%) who received no neoadjuvant treatment, 2,042 (6.1%) who received neoadjuvant chemotherapy alone, 334 (1.0%) who received neoadjuvant radiation alone, and 14,614 (43.8%) who received both neoadjuvant chemotherapy and radiation therapy. Following exclusion criteria, 12,460 patients were included in the study population who received both neoadjuvant chemotherapy and radiation therapy (Figure 1). There were 11,880 patients (95.3%) who received nCRT, as defined by receiving radiation treatment starting within 6 weeks of their chemotherapy start date. There were 580 patients (4.7%) who received IC + nCRT, as defined by radiation treatment starting greater than 6 weeks after their chemotherapy start date. The utilization of IC + nCRT was consistent throughout the entire study period with the IC + nCRT group representing 4.6% of the population in the early time period (2006–2012) and 4.7% of the population in the late time period (2013–2015).

Figure 1.

STROBE diagram of the study population.

The mean age of the study population was 62.1 years (SD 9.4) and consisted of 87.9% males. There were 8,264 patients (66.3%) with tumors located in the esophagus and 3,453 patients (27.7%) with tumors located in the GE junction. There were 8,241 patients (66.1%) with clinical T3 disease and 7,228 patients (58.0%) with clinical node positive disease (Table 1). Prior to propensity score weighting, patient sex, race, Charlson/Deyo comorbidity scores, tumor location, clinical T stage, and histologic grade were not statistically different between patients who received IC + nCRT versus nCRT. Patients who received IC + nCRT were younger (mean age: 60.7 versus 62.2 years, p < 0.001), had more advanced clinical N stage (N2–N3: 14.7% versus 10.1%, p < 0.001), and were more often treated at academic/research facilities (67.1% versus 59.5%, p = 0.001). These covariates were balanced between groups following stabilized inverse probability of treatment weighting (Table 2).

Table 2.

Patient, tumor, and treatment characteristics stratified by treatment group (nCRT versus IC + nCRT) before and after propensity weighting.

	Before Propensity Weighting					After Propensity Weighting
	Treatment group					Treatment group
	nCRT		IC + nCRT		p-value	nCRT		IC + nCRT		p-value
	N	%	N	%		N	%	N	%
Sex
Male	10,434	87.83	515	88.79	0.487	10,439	87.87	509	87.77	0.948
Female	1,446	12.17	65	11.21		1,441	12.13	71	12.23
Age (mean, SD)	62.2	9.4	60.7	9.9	<0.001	62.1	9.4	62.1	9.5	0.968
Race
White	11,427	96.19	552	95.17	0.115	11,421	96.14	559	96.46	0.899
Black	228	1.92	10	1.72		227	1.91	10	1.77
Other	225	1.89	18	3.10		232	1.95	10	1.76
Charlson/Deyo Score
0	8,289	69.77	415	71.55	0.571	8,299	69.86	408	70.39	0.950
1	2,785	23.44	131	22.59		2,780	23.40	132	22.79
2–3	806	6.78	34	5.86		801	6.74	40	6.82
Tumor Location
Esophagus	7,880	66.33	384	66.21	0.548	7,879	66.32	383	66.03	0.989
GE Junction	3,286	27.66	167	28.79		3,292	27.71	162	27.89
Unknown	714	6.01	29	5.00		708	5.96	35	6.08
Clinical T Stage
T1	583	4.91	28	4.83	0.122	582	4.90	26	4.50	0.977
T2	2,121	17.85	90	15.52		2,108	17.75	104	17.95
T3	7,832	65.93	409	70.52		7,857	66.14	385	66.35
Unknown	1,344	11.31	53	9.14		1,332	11.21	65	11.19
Clinical N Stage
N0	4,121	34.69	163	28.10	<0.001	4,085	34.38	197	34.09	0.981
N1	5,650	47.56	292	50.34		5,666	47.69	280	47.99
N2–N3	1,201	10.11	85	14.66		1,226	10.32	61	10.47
Unknown	908	7.64	40	6.90		904	7.61	43	7.45
Histologic Grade
Well-Moderately Differentiated	4,560	38.38	219	37.76	0.639	4,556	38.35	220	38.01	0.972
Poorly Differentiated	5,663	47.67	272	46.90		5,659	47.63	276	47.65
Unknown	1,657	13.95	89	15.34		1,665	14.01	83	14.34
Facility Type
Community	213	1.79	4	0.69	0.001	207	1.74	8	1.38	0.951
Comprehensive	2,957	24.89	112	19.31		2,926	24.63	144	24.89
Academic/Research	7,068	59.49	389	67.07		7,110	59.85	347	59.94
Integrated	1,642	13.82	75	12.93		1,637	13.78	80	13.79

Overall Survival

The median follow-up for the study population was 1.75 years (0.78–3.5 years) and the overall median survival was 2.48 years. On univariate analysis before propensity weighting, patients who received IC + nCRT had significantly longer OS than patients who received nCRT (median OS: 3.38 years versus 2.45 years, p = 0.002).

After propensity weighting, patients who received IC + nCRT had significantly longer OS than patients who received nCRT (median OS: 3.38 years versus 2.45 years, p = 0.001). Figure 2 shows the propensity weighted Kaplan-Meier plots for OS by treatment group. The adjusted HRs for OS from the propensity score weighted Cox-proportional hazards model are shown in Table 3. Receipt of IC + nCRT was independently associated with improved OS compared to nCRT (HR 0.83; 95% CI 0.74–0.93; p = 0.001). Additional factors associated with improved OS included younger age, female sex, fewer comorbidities, lower clinical T and N stages, well-moderately differentiated histologic grade, and receipt of treatment at an academic/research or integrated facility (Figure 3).

Figure 2.

Propensity-weighted Kaplan-Meier overall survival curves comparing patients who received IC + nCRT versus nCRT. P-value from propensity weighted univariate Cox proportional-hazard model.

Table 3:

Association between patient, tumor, and treatment characteristics and survival using propensity weighted Cox Proportional Hazards regression.

	HR	95% CI	P-value
Treatment group
nCRT	Ref.		0.001
IC + nCRT	0.83	[0.74, 0.93]
Sex
Male	Ref.		<0.001
Female	0.86	[0.80, 0.92]
Age	1.01	[1.01, 1.02]	<0.001
Race
White	Ref.		0.376
Black	1.02	[0.86, 1.23]
Other	0.88	[0.72, 1.05]
Charlson/Deyo Score
0	Ref.		<0.001
1	1.11	[1.04, 1.17]
2–3	1.18	[1.08, 1.30]
Tumor Location
Esophagus	Ref.		0.066
GE Junction	0.99	[0.93,1.04]
Unspecified	1.12	[1.02,1.23]
Clinical T Stage
T1	Ref.		<0.001
T2	1.00	[0.88, 1.12]
T3	1.16	[1.04, 1.30]
Unknown	1.32	[1.15, 1.52]
Clinical N Stage
N0	Ref.		<0.001
N1	1.18	[1.12, 1.24]
N2–N3	1.37	[1.26, 1.50]
Unknown	1.05	[0.94, 1.18]
Histologic Grade
Well-Moderately Differentiated	Ref.		<0.001
Poorly Differentiated	1.27	[1.21, 1.34]
Unknown	0.93	[0.87, 1.01]
Facility Type
Community	Ref.		<0.001
Comprehensive	0.93	[0.78, 1.10]
Academic/Research	0.76	[0.64, 0.90]
Integrated Network	0.79	[0.66, 0.94]
Year of Diagnosis
2006–2012	Ref.		0.064
2013–2015	0.95	[0.90, 1.00]

Figure 3.

Forest plot of hazard ratios for overall survival from propensity weighted Cox-proportional hazards model.

Pathologic Complete Response

There were 9,707 patients from the initial study population who had pCR data available. Overall, 1,754 patients (18.0%) had a pCR following neoadjuvant therapy. On univariate analysis, patient and disease characteristics that were associated with pCR rate included female sex (female: 20.6%, male: 17.7%, p = 0.013), esophageal tumor location (esophageal: 20.2%, GEJ: 13.7%, p < 0.001), lower clinical T stage (T1 23.0%, T2 22.7%, T3 17.7%, p < 0.001), and year of diagnosis (2013–2015: 19.2%, 2006–2012: 16.9%, p = 0.004).

Following propensity weighting, on univariate analysis there was a trend towards higher pCR rates in patients who received IC + nCRT versus nCRT, although this did not reach statistical significance (IC + nCRT: 21.2%, nCRT: 16.9%, p = 0.091). When stratified by time period of diagnosis, on univariate analysis there was no difference in pCR rates in the early time period (2006–2012) between patients who received IC + nCRT versus nCRT (16.9% pCR rate versus 15.1%, p = 0.513). However, in the later time period (2013–2015), patients who received IC + nCRT had a significantly higher pCR rate versus patients who received nCRT (27.4% pCR rate versus 18.8%, p = 0.003). On multivariable analysis a two-way interaction was evaluated between treatment group and time period of diagnosis. This two-way interaction was statistically significant, demonstrating the effect of treatment group on pCR rate was significantly different in the early versus late time periods. For patients diagnosed from 2006–2012, there was no difference in the odds of pCR between patients that received IC + nCRT versus nCRT (OR 0.90; 95% CI 0.63–1.31; p = 0.601). For patients diagnosed from 2013–2015, IC + nCRT was associated with significantly higher odds of pCR compared to nCRT (OR 1.59; 95% CI 1.14–2.21; p = 0.007). Additional factors associated with pCR on multivariable analysis include female sex, older age, esophageal tumor location, and lower T stage (Figure 4).

Figure 4.

Forest plot of odds ratios for pathologic complete response from propensity weighted multivariable logistic regression.

DISCUSSION

Despite advances in treatment, esophageal cancer continues to have dismal outcomes with 5-year survival of 25–47% in those who were selected to undergo neoadjuvant therapy and surgery (3, 7, 22). While nCRT followed by surgery or perioperative chemotherapy have become the standard of care in patients who have operable disease, the optimal neoadjuvant treatment regimen for these patients remains under debate. Given that most patients die of systemic disease, and face significant risk of locoregional recurrence with either nCRT or perioperative chemotherapy, it is plausible that patients may benefit from a combination of intensified systemic chemotherapy in combination with preoperative CRT. The addition of IC to nCRT treatment regimens could improve systemic control while maintaining the local benefits of CRT. With 12,460 patients, this study is the largest cohort study to date using a nationwide database to compare treatment outcomes of IC plus nCRT versus nCRT alone for esophageal and GEJ adenocarcinoma, where sequential timing of chemotherapy before radiation was used as a surrogate for the use of IC prior to nCRT.

Several single institution retrospective studies have shown that the addition of IC to nCRT for resectable esophageal cancer is associated with improved pCR rates, OS, and DFS (7–9). The utility of the addition of IC to nCRT was subsequently evaluated in a phase II randomized clinical trial (NCT 00525915) of 126 patients with resectable esophageal cancer receiving trimodal therapy who were randomized to either nCRT or IC plus nCRT between 2005–2013. The chemotherapy regimen used was oxaliplatin/5-FU for both IC and nCRT phases. The primary outcome of the trial was pCR with secondary outcomes including OS, R0 resection rate and post-operative mortality. The trial observed a pCR rate of 26% in patients in the IC arm versus 13% in the non-IC arm, with a p-value of 0.094 (15). Although this did not meet the threshold of statistical significance with an alpha of < 0.05, it did show a trend in favor of the use of IC, raising the question of whether the study may have been underpowered to detect a statistically significant difference in pCR. Additionally they did not find a statistically significant difference in OS between the two groups—with a median OS of 45.6 months (3.8 years) in the non-IC arm versus 43.7 months (3.6 years) in the IC arm—and concluded that further development of IC before nCRT would not be beneficial (15). However, a follow-up subgroup analysis found that patients with moderate or well differentiated tumors did have higher 5-year OS in the IC group (74%) versus non-IC group (50%) and suggested that perhaps there might be a role for IC in this subset of patients (16).

In our study, which spanned 2006–2015, we similarly observed an overall trend towards improved pCR rates with IC + CRT that did not reach statistical significance on univariate analysis (21.2% versus 16.9%). However, further analysis found that while patients diagnosed in the early time period (2006–2012) did not have higher odds of pCR with IC + nCRT compared to those who had nCRT alone (16.9% versus 15.1%), in the later time period (2013–2015), IC + nCRT was associated with significantly higher odds of pCR (27.4% versus 18.8%). This time cutoff is relevant because it coincides with the publication of the landmark CROSS trial, which established carboplatin/taxol as the standard chemotherapy regimen used during nCRT as opposed to earlier cisplatin-based nCRT regimens (3). In our current study, there was also a statistically significant survival benefit with IC + nCRT, with median OS of 3.38 years in the IC + nCRT group versus 2.45 years in the nCRT group.

There are several potential reasons for why these results favor the use of IC + nCRT despite the results from the only phase II randomized clinical trial (NCT 00525915) demonstrating no survival benefit with the use of IC amongst all patients. First, the study populations differed in that this current study only included adenocarcinoma patients and represents national level data. The NCT 005259155 trial also had a slightly higher percentage of patients with GEJ tumor (33% vs 28%). The study populations were similar in terms of percent of patients with baseline node positive disease (>60%) and percent of patients with poorly differentiated tumors (~50%) (15). Additionally, using data from the NCDB, we were unable to capture patients that were started on neoadjuvant therapy and subsequently had disease progression or complications that prevented them from progressing to definitive surgical resection. Given the longer time-course involved with IC prior to nCRT, it is possible that patients with more aggressive biology demonstrated progression during this longer time-course, which may bias these results. In the NCT 00525915 trial, only 1 patient out of 63 patients developed M1 disease during the IC chemotherapy phase of the trial. Following nCRT, 4 patients in the IC arm did not proceed to surgery due to M1 disease versus 2 out of 63 in the non-IC arm suggesting this effect may be small (15).

IC may also play a role in prognostication as well as helping to personalize the choice of chemotherapy during nCRT. The CALGB 80803 (Alliance) trial used PET/CT to assess the response to IC in patients with esophageal cancer receiving either a modified FOLFOX-6 or carboplatin/taxol regimen. Following induction therapy, those patients that showed a response based on PET/CT continued with the same chemotherapy combination during the CRT phase, while non-responders received the alternate regimen during their CRT phase. Patients who were started on FOLFOX-6 and were PET responders had a pCR rate of 38% and patients who were started on carboplatin/taxol who were PET responders had a pCR rate of 10.75%. Among non-responders who switched to the alternate regimen based on PET imaging, the pCR rate was 15.6% which compared favorably to a historical control pCR rate of 5% (12). A positive response to IC based on PET/CT findings was also statistically correlated with OS, suggesting that such a response could also be used as a prognostic indicator (13). IC can thus help tailor patients’ nCRT regimen by identifying responders to a given regimen and future trials evaluating the efficacy of IC should incorporate this approach. In the NCT 00525915 trial, oxaliplatin/5-FU was used for both IC and CRT phases and did not allow for response based alterations in chemotherapy regimens during CRT which may limit the clinical benefit of IC (15). Moreover, as systemic chemotherapy regimens for esophageal cancer continue to evolve and improve (e.g. FOLFOX-6, FLOT), the impact of induction chemotherapy may also increase.

Our study has several limitations, including the retrospective nature of the study, which is inherently subject to higher risk of bias and confounding than a prospective randomized trial. We have taken several steps to minimize these factors in our study, including propensity matching the cohorts so that covariates are evenly distributed between the treatment groups, assessing for potential interactions between covariates, and using logistic regression to evaluate treatment effects. There are also limitations related to the use of a large national database, such as the potential for miscoding and variations in treatment details across different hospitals, as well as certain limitations of the NCDB. One such limitation is the lack of data on specific chemotherapy regimens and number of cycles for either group. It is possible that the improved results in the IC + nCRT group is due to better chemotherapy regimens compared with the nCRT group, though we do not have any reason to believe this would be the case.

Notably, we used time interval from chemotherapy start date relative to radiotherapy start date as a surrogate for induction chemotherapy before nCRT versus nCRT. Due the nature of the database, we cannot know for sure whether all patients who started radiation therapy greater than six weeks after chemotherapy truly were intended to receive IC, or whether there was another reason for the selection and timing of treatment. One possible alternative explanation for sequential timing could include patients who were initially either erroneously thought to have metastatic disease or had equivocal findings suggestive of metastatic disease (e.g. a nonspecific pulmonary nodule) and thus were initiated on a chemotherapy regimen to assess their response to systemic therapy. Another hypothetical alternative could include patients who initially refused surgery, and thus were being treated with definitive chemotherapy. Another limitation is that patients who were started on a neoadjuvant regimen whose disease progressed during this treatment and thus did not undergo surgical resection are not accounted for in this study, which may overestimate the OS and pCR rates in our study.

Despite these limitations, our study has several strengths. The use of a national database allows for a large sample size, as well as data about survival as well as pCR. It also offers an assessment based on real world outcomes from a broad array of hospital environments rather than the controlled environment of a clinical trial.

In conclusion, this NCDB study suggests that IC before nCRT for esophageal and GEJ adenocarcinoma is associated with an overall survival benefit and improvement in pCR rates. This strategy has potential to improve upon the current standard of care and warrants further investigation with standardized chemotherapy and chemoradiation regimens via a randomized controlled trial.

SYNOPSIS.

In this retrospective study, use of induction chemotherapy before neoadjuvant chemoradiation therapy (nCRT) is associated with a significant survival benefit as well as higher pathologic complete response rates when compared to nCRT without induction chemotherapy for esophageal and gastroesophageal adenocarcinoma.