Phase 2 trial of preoperative irinotecan plus cisplatin and conformal radiotherapy, followed by surgery for esophageal cancer


  • Jennifer J. Knox MD,

    Corresponding author
    1. Department of Medical Oncology, Princess Margaret and Toronto General Hospitals of the University Health Network, Toronto, Ontario, Canada
    • Department of Medical Oncology, Princess Margaret Hospital, 610 University Avenue, 5-210, Toronto, ON, M5G 2M9 Canada
    Search for more papers by this author
    • Fax: (416) 946-6546

  • Rebecca Wong MD,

    1. Department of Radiation Oncology, Princess Margaret and Toronto General Hospitals of the University Health Network, Toronto, Ontario, Canada
    Search for more papers by this author
  • Antonio L. Visbal MD,

    1. Department of Medical Oncology, Princess Margaret and Toronto General Hospitals of the University Health Network, Toronto, Ontario, Canada
    Search for more papers by this author
  • Anne M. Horgan MD,

    1. Department of Medical Oncology, Princess Margaret and Toronto General Hospitals of the University Health Network, Toronto, Ontario, Canada
    Search for more papers by this author
  • Maha Guindi MD,

    1. Department of Pathology, Princess Margaret and Toronto General Hospitals of the University Health Network, Toronto, Ontario, Canada
    Search for more papers by this author
  • Jennifer Hornby BSc,

    1. Division of Thoracic Surgery, Department of Surgical Oncology, Princess Margaret and Toronto General Hospitals of the University Health Network, Toronto, Ontario, Canada
    Search for more papers by this author
  • Wei Xu PhD,

    1. Department of Biostatistics, Princess Margaret and Toronto General Hospitals of the University Health Network, Toronto, Ontario, Canada
    Search for more papers by this author
  • Jolie Ringash MD,

    1. Department of Radiation Oncology, Princess Margaret and Toronto General Hospitals of the University Health Network, Toronto, Ontario, Canada
    Search for more papers by this author
  • Shaf Keshavjee MD,

    1. Division of Thoracic Surgery, Department of Surgical Oncology, Princess Margaret and Toronto General Hospitals of the University Health Network, Toronto, Ontario, Canada
    Search for more papers by this author
  • Eric Chen MD,

    1. Department of Medical Oncology, Princess Margaret and Toronto General Hospitals of the University Health Network, Toronto, Ontario, Canada
    Search for more papers by this author
  • Masoom Haider MD,

    1. Department of Radiology, Princess Margaret and Toronto General Hospitals of the University Health Network, Toronto, Ontario, Canada
    Search for more papers by this author
  • Gail Darling MD

    1. Division of Thoracic Surgery, Department of Surgical Oncology, Princess Margaret and Toronto General Hospitals of the University Health Network, Toronto, Ontario, Canada
    Search for more papers by this author

  • We gratefully acknowledge the support of Pfizer Canada.



Esophagectomy for locally advanced esophageal cancer (LAEC) is associated with limited survival. Trimodality therapy yields a small survival advantage, with cisplatin and 5-fluorouracil regimens most frequently studied. Newer regimens may impact these poor outcomes. This phase 2 trial assessed the feasibility and efficacy of induction chemoradiotherapy with cisplatin and irinotecan followed by esophagectomy.


Patients with LAEC of the thoracic esophagus or gastroesophageal junction underwent chemotherapy with preoperative irinotecan (65 mg/m2) plus cisplatin (30 mg/m2) on Weeks 1, 2, 4, 5, 7, and 8 with concurrent conformal radiotherapy (40 grays [Gy]/20 fractions during Weeks 4-7) and external beam boost (10 Gy/5 fractions at Week 8). Esophagectomy was performed between Weeks 12 and 16. Pathologic response was the primary endpoint with follow-up data on progression, survival, and toxicity as secondary endpoints.


Fifty-two patients were enrolled from November 2002 to October 2005. Nineteen patients had American Joint Committee on Cancer stage II, 22 had stage III, and 11 had stage IVA disease. Grade 3 to 4 toxicity (graded according to the National Cancer Institute Common Toxicity Criteria 2.0) during induction included neutropenia (36%), febrile neutropenia (8%), diarrhea (10%), and esophagitis (4%). Three patients withdrew from treatment due to toxicity. There was 1 treatment-related death. Clinical responses included complete response in 2%, partial response in 30%, stable disease in 62%, and progressive disease in 6% of patients. Dysphagia improved/resolved in 72% of patients during induction. Forty-three patients underwent esophagectomy and 7 (16%) achieved pathologic complete responses. Median and 3-year overall survival for patients receiving trimodality therapy was 36 months and 51%, respectively.


In LAEC, concurrent irinotecan/cisplatin and radiotherapy followed by esophagectomy is reported to be associated with dysphagia improvement in 72% of patients, a significant but manageable toxicity profile, and encouraging survival compared with historic controls. Cancer 2010. © 2010 American Cancer Society.

Cancer of the esophagus has a poor prognosis. Esophagectomy for locally advanced esophageal cancer (LAEC) yields limited survival. For patients with resectable tumors, surgical series report 5-year survival rates of 22% to 25%.1, 2 Preoperative chemoradiotherapy has shown promise. Standard regimens of cisplatin and 5-fluorouracil (5-FU) have been extensively studied in small randomized trials,3-5 with most reported as negative studies. Several meta-analyses suggest there is a small survival advantage to trimodality therapy with an overall hazard ratio (HR) of approximately 0.86.6-8 More recently, the Cancer and Leukemia Group B (CALGB) 9781 trial (n = 56) reported a survival advantage with trimodality therapy compared with surgery alone after a median of 6 years of follow-up (median survivals of 4.5 vs 1.8 years, respectively; P = 0.002).9

In an effort to improve on these outcomes, newer combinations of chemotherapy with radiotherapy have been evaluated. Both irinotecan and cisplatin individually have activity against carcinoma of the esophagus. Both agents have radiosensitizing properties and are well tolerated in this patient population. Irinotecan, a relatively novel agent in the context of esophageal cancer, has shown good response rates among patients with advanced disease10 and is being further studied in the management of potentially resectable LAEC. In a phase 1 study of patients with stage II to III esophageal cancer (n = 19), irinotecan was combined with cisplatin and radiotherapy preoperatively.11 This novel combination appeared to be active and well tolerated. Therefore, we designed a phase 2 trial to further assess feasibility and efficacy of induction chemoradiotherapy with irinotecan plus cisplatin followed by surgery for patients with potentially resectable LAEC.


This was a single-arm phase 2 trial conducted at the Princess Margaret and Toronto General Hospitals in Toronto, Ontario, Canada. The study was approved by the Institutional Review Board, and patients were required to provide signed informed consent.


Patients with histologic proof of squamous cell carcinoma or adenocarcinoma of the thoracic esophagus or gastroesophageal junction (GEJ); clinical T1NI, T2 to 4 N0 to 1, and M0 to M1A disease according to the American Joint Committee on Cancer (AJCC); and an Eastern Cooperative Oncology Group (ECOG) performance status of 0 to 2 were eligible. M1A patients were included for tumors of the GEJ, if the celiac lymph nodes were in proximity to the primary tumor. Patients were staged with computed tomography (CT) scan of the chest and abdomen and endoscopy, with endoscopic ultrasound used selectively. At endoscopy, clips were placed to mark the upper and lower extent of the tumor to aid in radiation planning. Fluorodeoxyglucose (18F-FDG) positron emission tomography/CT (PET/CT) was not routine but was performed in the latter half of the study on 18 patients. Adequate baseline laboratory parameters including a neutrophil count ≥1.5 × 109/L; a platelet count ≥100 × 109/L; total bilirubin ≤1.25× the upper limit of normal (ULN); aspartate transaminase (serum glutamic oxaloacetic transaminase), alkaline phosphatase, and alanine transaminase (serum glutamic-pyruvic transaminase) ≤2.5 × ULN; serum creatinine ≤160 μmol/L; or creatinine clearance ≥60 mL/min were required.


Patients received cisplatin at a dose of 30 mg/m2 plus irinotecan at a dose of 65 mg/m2 once weekly on Weeks 1 2, 4 5, 7, and 8, with concurrent radiotherapy during Weeks 4 to 8. Adequate hydration, antiemetics, and supportive medications (eg, atropine, loperamide) were administered per institutional policy. The induction regimen was 1 cycle (2 doses) shorter than the phase 2 schedule by Ilson et al10 in an effort to shorten the overall preoperative treatment time. Chemotherapy was delayed for toxicities until recovery to ≤ grade 1 and/or dose reduced for ≥ grade 3 nonhematologic toxicity. Grade 3 neutropenia on treatment day resulted in a delay and an irinotecan dose reduction of 15 mg/m2. Grade 4 neutropenia on treatment day resulted in a delay and a 20 mg/m2 dose reduction. Irinotecan was reduced by 10 mg/m2 for a platelet count of 50 to 100 × 109/L and by 15 mg/m2 for a platelet count <50 × 109/L. Febrile neutropenia on any day resulted in a delay if necessary and a decrease in irinotecan by 20 mg/m2 and cisplatin by 10 mg/m2. Granulocyte–colony-stimulating factor (G-CSF) was not used outside of the occurrence of febrile neutropenia. Elevations of creatinine to grade 2 to 3 resulted in a cisplatin dose reduction of 10 mg/m2. A grade 4 increase in creatinine resulted in discontinuation of cisplatin.


Patients underwent 3-dimensional (3D) CT planning for all phases of external beam radiotherapy. The dose was 40 grays (Gy) in 20 fractions delivered typically with a 6-field conformal technique (Weeks 4-7). This was followed by an external beam conformal boost (Week 8) at a dose of 10 Gy in 5 fractions to take the total dose up to 50 Gy in 25 fractions (Fig. 1). The clinical target volume (CTV) was defined as the primary tumor with a margin of 3 to 4 cm in the cranial-caudad directions for the primary (and a minimum of 1 cm for the lymph nodes), and a 1-cm margin (modified for anatomic boundaries) circumferentially. For the boost volume, the CTV margin could be reduced to 2 cm in the cranial-caudad direction. The celiac axis was included for distal esophagus and GEJ tumors wherever possible. The planning target volume (PTV) was created using a margin of 1 cm superiorly and inferiorly, and 0.5 cm circumferentially. The plan was optimized to cover the PTV with 95% while accepting dose heterogeneity of ±5%. The maximum spinal cord dose permitted was 45 Gy. For the lung, we used both a mean lung dose of <18 Gy and the total lung volume receiving >20 Gy was <30% (V20 of <30%). The volume of the heart receiving >45 Gy was <60% (V45 of <60%) and for the liver, we used a mean dose of <30 Gy and the volume of the liver receiving >30 Gy was <60% (V30 of <60%).

Figure 1.

The trial schema is shown.


At the completion of induction therapy, if restaging confirmed no evidence of distant disease, patients underwent esophagectomy approximately 4 to 8 weeks after treatment. The surgical technique was comprised of a McKeown esophagectomy for lesions localized in the middle third of the esophagus and either an Ivor-Lewis resection or left thoracoabdominal approach with a left neck incision for lesions in the lower third of the esophagus and/or GEJ. Transhiatal esophagectomy was permitted but not encouraged. En bloc resection of overlying mediastinal pleura, a cuff of the diaphragmatic crura, and mediastinal fat was recommended. Recommended lymphadenectomy included lower mediastinal lymph nodes (Stations 7, 8, and 9), and abdominal lymph nodes (Stations 15, 16, 17, 18, 19, and 20). For tumors of the mid-esophagus, resection of Stations 2 and 4R was also recommended. Proximal and distal margins, measured in the operating room, were >5 cm beyond the macroscopic tumor.


The tumor was submitted in its entirety for histologic examination by 1 pathologist (M.G.). Tumor regression was graded on a scale of 1 to 5 based on the presence of residual tumor cells and extent of fibrosis, as described by Mandard et al.12 Grade 1 (G1) was complete pathologic response demonstrated by the absence of residual tumor and fibrosis extending through the layers of the esophageal wall. Grade 2 (G2) was minimal residual disease characterized by rare residual tumor scattered throughout the fibrosis. Grade 3 (G3) was characterized by an increase in the number of residual tumor cells but with predominance of fibrosis. Grade 4 (G4) was residual tumor outgrowing the fibrosis and grade 5 (G5) was the absence of any tumor regression. Cytologic atypia was classified into mild, moderate, and marked. The inflammatory reaction was graded as grade 0 9absent or mild), grade 1 (moderate), and grade 2 (extensive).

Evaluation and Toxicity Criteria

During preoperative treatment, patients were assessed at least weekly. Toxicity was graded according to the National Cancer Institute Common Toxicity Criteria, version 2.0 (NCI CTC v 2.0).

Dysphagia was measured as a patient-reported outcome using the NCI CTC v 2.0 at baseline and weekly during preoperative therapy. The self-administered quality of life (QoL) tool Functional Assessment of Cancer Therapy-Esophagus (FACT-E) was completed at baseline; at Weeks 7, 12 to 14 (presurgery); at 1, 3, and 12 months after resection;and then yearly until Year 5. FACT-E evaluates the effect of treatment on QoL using an esophageal cancer subscale (17-item questionnaire) in combination with the Functional Assessment of Cancer Therapy-General (FACT-G) questionnaire, which includes 27 items.13

Statistical Analysis

This study was designed to assess the feasibility, efficacy, and safety of induction chemoradiotherapy followed by surgery. The primary outcome of interest was pathological complete response (pCR) rate. Secondary outcomes included clinical response, adverse effects, dysphagia relief, QoL and overall survival (OS). We were interested in a well-tolerated regimen with pCRs >20%. The planned sample size was 36 non-M1A patients taking into account both the response rate we anticipated (20%) and acute toxicities we wanted to avoid as per Bryant and Day.14 M1A patients were not excluded but we wanted a large enough non-M1A population for evaluation of survival outcomes with this trimodality approach to compare with other studies that may exclude M1A patients. The design provided a type I error rate of 10% for response and 10% for toxicity. The overall type I error rate was 1% (2-sided α = 0.01), with planned power of 85% (β = 0.15). A planned interim analysis, after 13 patients were recruited, demonstrated more than 2 pCRs, and no acute (from Week 1 to 12) grade 3 nonhematologic toxicities, or prolonged grade 4 hematologic toxicities; thus, the study was continued to completion.

Clinical and radiologic responses were described by Response Evaluation Criteria In Solid Tumors (RECIST) criteria.15 The start date for time to recurrence, relapse-free survival (RFS), and OS was the first day of preoperative chemotherapy. Proportions, mean values with 95% confidence intervals (95% CI), and medians were used to describe patient characteristics and outcomes where appropriate. The Kaplan-Meier approach was used to estimate survival rate.

Patients were evaluable for toxicity (Evaluableacute tox) or evaluable for clinical response (EvaluableCR) if they received any of the preoperative study interventions and continued consent. For clinical response, repeat CT assessment was also required. Patients were evaluable for pathologic response (Evaluabletrimodality) if they received at least 75% of the planned preoperative chemotherapy (allowing appropriate dose reductions) and radiotherapy, and underwent surgery. Patients were evaluable for surgical morbidity (Evaluablesurgical morbidity) if they underwent surgery.


From November 2002 to October 2005, 52 patients were enrolled and commenced therapy (Evaluableacute tox). The mean age was 60 years (range, 33-79 years), and 77% were male. The tumor was located in the GEJ in 15 patients (29%) and in the thoracic esophagus in 37 (71%). Histology was adenocarcinoma in 37 (71%), squamous cell carcinoma in 13 (25%), and poorly differentiated large cell carcinoma in 2 patients (4%). Using clinical staging, 11 patients were classified with T2 disease, 25 with T3 disease, and 6 with T4 disease (by virtue of extraesophageal extension into the mediastinal soft tissues, but not the adjacent viscera); 15 patients were classified with N0 disease, and 37 with N1 disease; 41 patients were classified with M0 disease, and 11 as having M1A disease. The clinical stage at enrollment was IIA in 13 patients (25%), IIB in 6 patients (12%), III in 22 patients (42%), and IVA in 11 patients (21%).

Of the 52 enrolled patients, 3 patients withdrew from treatment due to toxicity and proceeded directly to surgery (Fig. 2). Two refused further treatment during induction and did not undergo surgery. There were 2 preoperative deaths, 1 patient died of a stroke after 1 cycle of chemotherapy (possibly related to chemotherapy), and 1 patient died of central line sepsis, which occurred 6 months from chemotherapy and was deemed not likely related to the treatment. Preoperative therapy was completed in 45 patients (EvaluableCR); however, 2 patients developed disease progression while receiving on treatment and were not treated surgically. The remaining 43 patients proceeded to surgery (Evaluabletrimodality) (Fig. 2). Pathologic response data were available on 43 patients.

Figure 2.

Patient disposition during study is shown. EvaluableCR indicates evaluable for clinical response; Evaluabletrimodality, evaluable for pathologic response.

Dysphagia Relief

At diagnosis, 47 of 52 (90%) patients presented with dysphagia. Thirty-four of 47 (72%) patients reported their dysphagia improved or resolved: 22 of 47 (47%) after 1 cycle of chemotherapy and 12 of 47 (25%) after 2 cycles. No change was reported by 5 patients (11%), and 6 patients (13%) reported worsening of dysphagia. Response could not be evaluated in 1 patient (2%) and 1 patient (2%) had a mixed response. In total, only 6 patients required feeding tubes, 2 of which were inserted before treatment because of poor nutritional status and 4 during chemoradiation.


Preoperative chemoradiation was found to have a statistically significant detrimental effect on health-related QoL (FACT-E) scores during treatment (mean score declined from 127 at baseline to 120 at 7 weeks after treatment), but scores recovered to baseline before surgery. Health-related QoL decreased again after surgical intervention (mean 115, 1 month postoperatively) but returned to baseline levels within 3 months. Significant improvement over baseline was observed at 1 year from surgery in 28 patients who continued to comply with measurement (mean, 139). Complete data were reported previously.16

Acute Toxicity

Table 1 lists all adverse events encountered during the preoperative treatment, considered possibly related to the preoperative treatment. Toxicity was moderate overall and included the following: grade 3 to 4 neutropenia (36%), fatigue (16%), esophagitis/dysphagia (16%), and anorexia (14%). Febrile neutropenia occurred in 4 patients (8%). Pulmonary emboli (PE) occurred in 3 patients (6%) while receiving treatment, 1 of which was asymptomatic and found on preoperative restaging. Because PEs are common in this patient population, it is unclear whether this protocol causes an increase in PE rate over baseline.

Table 1. Chemoradiotherapy Toxicitya
N=52Grade 2Grade 3Grade 4Grade 5
No. (%)No. (%)No. (%)No. (%)
  • a

    Toxicity was graded according to the National Cancer Institute Common Toxicity Criteria, version 2.0.

 Anemia17 (33)2 (4)1 (2) 
 Leukopenia16 (31)20 (38)4 (8) 
 Neutropenia17 (33)12 (23)7 (13) 
 Febrile neutropenia4 (8)  
 Nausea22 (42)3 (6)  
 Anorexia14 (27)4 (8)3 (6) 
 Esophagitis15 (29)2 (4)  
 Diarrhea7 (13)5 (10)  
 Constipation16 (31)4 (8)  
 Vomiting7 (13)2 (4)  
 Dysphagia4 (8)1 (2)5 (10) 
 Cramping2 (4)   
 Fatigue23 (44)6 (12)2 (4) 
 Pain9 (17)1 (2)  
 Alopecia10 (19)   
 Dehydration7 (13)4 (8)  
 Infection3 (6)4 (8)  
 Pulmonary embolus 3 (6) 
 Candidiasis3 (6)   
 Pneumonia3 (6)  
 Neuropathy1 (2)  
 Perforation1 (2)  
 Stroke   1 (2)


Significant down-staging was found both clinically and pathologically. Radiologic (clinical) response by RECIST criteria was noted in 36% of the 45 EvaluableCR patients, stable disease was observed in 60%, and progressive disease was noted in 4%. Table 2 demonstrates details of down-staging. The pathologic tumor regression by the Mandard method was 55% in the 43 Evaluabletrimodality patients. pCR (G1) was achieved in 7 of 43 (16%), minimal residual disease (G2) in 16 of 43 (37%), G3 in 15 of 43 (35%), and G4 in 5 of 43 (12%) patients.

Table 2. Downstaging After Preoperative Therapy
Stage45 EvaluableCR Patients43 Evaluabletrimodality Patients
DescriptorClinical StageClinical StagePathological Stage
At EnrolmentAfter InductionAt EnrolmentAfter InductionPostresection
  1. EvaluableCR indicates evaluable for clinical response; Evaluabletrimodality, evaluable for pathological response.

0 1 (2%) 1 (2%)7 (16%)
I    4 (9%)
IIA9 (20%)28 (62%)9 (21%)28 (65%)11 (26%)
IIB5 (11%)1 (2%)5 (11%)1 (2%)7 (16%)
III21 (47%)9 (20%)20 (47%)8 (19%)14 (33%)
IVA10 (22%)4 (9%)9 (21%)3 (7%) 
IVB 2 (4%) 2 (5%) 

Surgical Outcome

Forty-six patients underwent surgery (Evaluablesurgical morbidity), 43 of whom completed the induction regimen. Surgical approaches included left thoracoabdominal in 21 patients (19 with anastomoses in the left neck and 2 in the chest), McKeown in 8, Ivor-Lewis in 8, transhiatal in 7, and a minimally invasive approach in 2 patients (thoracoscopic mobilization of the esophagus followed by laparoscopic abdominal portion converted to open laparotomy due to multiple adhesions in 1 patient and transhiatal laparoscopic resection in 1 patient, both with cervical anastomoses). The median hospital stay was 14 days (range, 8-96 days). Postoperative complications in 46 Evaluablesurgical morbidity patients included: anastomotic leak (22%), atrial fibrillation (24%), pneumonia (22%), and delirium and agitation (12%). There were 3 postoperative deaths, 1 at 4 weeks after surgery due to sepsis from a multidrug resistant pneumonia, 1 at 8 weeks due to aspiration, and 1 at 22 weeks after a lengthy postoperative course due to complications including sepsis and ultimately documented recurrent cancer. Their pathologic responses were G2, G1, and G3, respectively. Among the 43 patients evaluable for trimodality therapy, complete (R0) resection was achieved in 98%.


At a median follow-up of 30.2 months (range, 1.3 months - 5.8 years), 36 of 52 (69%) patients had died for a median survival of 30 months from the date of first treatment, and a 3-year survival probability of 46.2%. Of the 45 EvaluableCR patients, 30 patients had died, with a median survival of 32 months from the date of receiving the first treatment on study, and a 3-year survival probability of 48.9%. Of the 43 Evaluabletrimodality patients, 28 patients had died, with a median survival of 36 months and a 3-year OS rate of 51.1%. The median RFS was 17 months (from the surgery date), and the 3-year RFS rate was 34.2% (Fig. 3). Of the 29 patients who had developed disease recurrence at the time of last follow-up, in 21 (72%) the first recurrences were at distant sites only, in 3 (10%) the first recurrences were regional plus distant (in which regional is defined as mediastinal lymph nodes), in 3 (10%) the first recurrences were regional only, in 1 (3%) the first recurrence was regional plus local (local disease is defined as anastomotic recurrences), and in 1 (3%) the first recurrence was local disease only (Table 3). Sixteen patients received subsequent chemotherapy in the first-line metastatic setting, 3 received second-line therapies, and 3 received third-line therapies.

Figure 3.

Overall survival and relapse-free survival are shown for patients who were evaluable for pathologic response (Evaluabletrimodality) (n = 43).

Table 3. Patterns of First Recurrence to Date by Pathologic Regression
GenderAge, YearsSite of Primary TumorSite of First Disease RecurrenceTime to Disease Recurrence From Initiation of Therapy, Months
  1. LN indicates lymph nodes; GEJ, gastroesophageal junction.

Pathologic response: grade 1 (n=3 of 7)
 Female59ThoracicMediastinal LN27.4
 Male76ThoracicMediastinal/abdominal LN50.6
Pathologic response: grade 2 (n=12 of 16)
 Male64ThoracicPeritoneal carcinomatosis9.2
 Male57GEJMediastinal LN9.2
 Male67ThoracicRetroperitoneum, pleural effusion9.2
 Male59GEJS1/S2/L5 and peritoneal carcinomatosis9.3
 Male66ThoracicPeritoneal carcinomatosis11.0
 Male54ThoracicSupraclavicular LN12.7
 Male72GEJMediastinal/abdominal LN, ascites, pleural effusions13.9
 Male47GEJPrimary anastomosis and mediastinal LN17.1
 Male61GEJAbdominal LN/lung34.3
Pathologic response: grade 3 (n=11 of 15)
 Male52GEJBrain and liver7.2
 Male67GEJMediastinal/abdominal LN7.6
 Female33ThoracicPrimary anastomosis13.8
 Male61ThoracicAbdominal LN20.4
 Male69GEJAbdominal LN27.4
Pathologic response: grade 4 (n=3 of 5)
 Male49ThoracicMediastinal LN and subcutaneous16.1

We performed subgroup analysis based on pathologic response grade with respect to survival. When exploring the correlation between pathologic response and the risk of recurrence (Table 3), those who achieved the strictest definition of pCR were separated from all others. Of the 7 patients with pCR (G1), 3 had developed disease recurrence and 4 had died (3 from disease recurrence and 1 early postoperative death), with a 3-year RFS rate of 67% and an OS rate of 71%. Perhaps surprisingly, patients with pathologic response of G2 (minimal residual disease only) were found to have a similar risk of disease recurrence and survival compared with those with pathologic response G3 and G4 (P = .45). The median RFS was 50.5 months for patients with G1 response, 12.7 months for those with G2 response, and 20.4 months for those with a G3 to G4 response. Of 36 patients with pathologic response G2 to G4, 26 had developed disease recurrence at the time of last follow-up, and 24 had died, with a 3-year RFS rate of 32% and an OS rate of 49% (Fig. 4).

Figure 4.

(Top) Overall survival is shown for patients who were evaluable for pathologic response (Evaluabletrimodality) (n = 43) (complete pathologic response [G1] vs minimal residual disease or greater [G2-G4]). (Bottom) Relapse-free survival is shown for Evaluabletrimodality patients (n = 43) by pathologic response (G1 vs G2-G4).


Preoperative therapy with irinotecan/cisplatin and concurrent radiation proved feasible with moderate toxicity for this patient population. Dysphagia improved early and significantly in 72% of patients with this symptom, largely eliminating the need for feeding tubes or parenteral feeds during induction. This is a real advantage over combination 5-FU and cisplatin regimens, for which worsening dysphagia is common17, 18 and enteral/parenteral nutritional support is encouraged. The 36% rate of grade 3 to 4 neutropenia with 8% febrile neutropenia, as well as moderate rates of grade 3 to 4 esophagitis and other gastrointestinal toxicities, suggest this regimen appears to be less toxic than the combination of taxanes with 5-FU and platinum.18-20 However, given the overall toxicity pattern observed, patient selection is key and this trimodality approach may not be appropriate for all patients.

There is still discussion in the literature regarding whether induction therapy increases perioperative morbidity and mortality. Two meta-analyses suggested a trend toward improvement in survival associated with an increase in perioperative mortality after neoadjuvant chemoradiotherapy.21, 22 A more recent systematic review restricted to 6 randomized controlled trials, including only patients with resectable esophageal carcinoma, demonstrated a significant improvement in 3-year survival for patients treated with neoadjuvant chemoradiotherapy followed by surgery compared with surgery alone (odds ratio, 0.53; 95%CI, 0.31-0.92 [P = .025]).23 Although the anastomotic leak rate (22%) in the current series may be higher than reported for surgery alone, the perioperative morbidity and mortality was comparable to similar phase 2 trials18, 20 and historic controls.24

Local control was excellent, R0 resection was achieved in 98% and significant pathologic response was documented in 53% of evaluable patients. Pathologic down-staging has been reported to be a significant factor in determining survival,25 and pCR rates vary according to different regimens used by institutions. A meta-analysis of randomized trials comparing neoadjuvant chemoradiotherapy and surgery versus surgery alone demonstrated a 21% pCR rate21 for trimodality therapy. A retrospective series of patients from 3 institutions treated with induction chemoradiotherapy reported a pCR rate of 18%.26 In a phase 2 trial, induction chemoradiotherapy with paclitaxel and cisplatin resulted in a 25% pCR rate at the expense of increased toxicity.19 Combinations of lower doses of paclitaxel with carboplatin or carboplatin and 5-FU have led to reported pCR rates of 18% and 25%, respectively.20, 27 Review of the literature does not reveal a standardized method for evaluation of residual tumor in resection specimens after chemotherapy and radiotherapy; this may explain some of the differences in pCR rates reported, which may not reflect true differences in pathologic response.23 For example, van Meerten et al20 used a tumor regression classification protocol for evaluation of neoadjuvant therapy for non-small cell lung cancer.27 Meluch et al used their own institutional definition for pathologic evaluation, but applied it to routine pathology reports from outside institutions.18 Similar to Henry et al,19 we used tumor regression grade on a scale of 1 to 5, based on the presence of residual tumor cells and extent of fibrosis according to the method described by Mandard et al,12 previously described for the pathologic assessment of esophageal and rectal carcinoma.28 Importantly, our samples were systematically processed and reviewed by 1 pathologist, in an attempt to minimize interobserver variability. The variability in pathologic evaluation among trials and the thorough pathologic assessment in our trial possibly explains why, despite our lower pCR rate, the overall pathologic response, OS, and disease-free survival were similar to other recently reported phase 2 trials.20, 27 In fact, we confirmed a higher risk of recurrence for patients with evidence of only very minimal microscopic disease remaining (G2), despite evidence of marked tumor regression, when compared with the complete pathologic responders (G1s) (HR, 2.58; P = .14). These G2 tumors appear to have similar recurrence and survival outcomes to patients with less marked pathologic response or the G3 and G4 tumors (HR, 1.35; P = .45). Lastly, we defined an anticipated pCR rate of 20% in the study design. At 16%, we have clearly not met that level. This combination does not appear to advance the field by the criteria of improved pCR, no matter how confident we are in our estimates of that endpoint.

The results of the current study are consistent with other phase 1 and 2 trial data with neoadjuvant cisplatin/irinotecan combined with radiation. The ECOG 1201 trial compared preoperative weekly cisplatin/irinotecan with cisplatin/paclitaxel, both combined with radiation, in esophageal adenocarcinoma.29 The median OS was 34.9 months and 21 months for the irinotecan and paclitaxel arms, respectively. A pCR rate of 15% to 16% was reported. Although these results compare favorably with historic controls of larger series of patients treated with surgery alone (median and 3-year survivals of 36 months and 51% vs 23 months and 34-38% extrapolated from survival curves),1, 2 this interpretation is exploratory outside a randomized trial. A recent study used a similar chemoradiation strategy to that studied herein, as definitive therapy (without surgery) for patients with LAEC, primarily squamous cell carcinoma.30 This study (n = 43) was comprised of 23% of patients with stage 1 disease, 44% with stage 2 disease, and 33% with stage 3, a generally less advanced cohort than enrolled in our study. Despite good clinical response rates, the 1-year and 2-year survival rates were reported as 63% and 28%. Contrast these to our findings of 1-year and 2-year survival rates of 84% and 67%, respectively. Whereas it is problematic to compare phase 2 trials, this may suggest a superior outcome with the trimodality approach. Ultimately, phase 3 data are required to determine the benefit of this regimen in the neoadjuvant setting over surgery alone and the comparability of this combination to conventional chemotherapy or to alternative combinations.31, 32

There is discussion in the literature regarding whether M1A patients should be included in multimodality trials. Historically, these patients were excluded from studies,3, 33 or were not analyzed independently4, 5, 9 hence, to the best of our knowledge, there are no randomized data addressing the optimal treatment for these patients. Retrospective series report 3-year survival rates of 12% and 18%, respectively, in patients with M1A disease treated with trimodality therapy (n = 18)34 and surgery alone (n = 52).35 In our subgroup analysis by M classification, patients staged as IVA had worse survival than those not staged as IVA. However, in the stage IVA patients who were treated with trimodality therapy (n = 9), there was a remarkable clinical down-staging, achieving notable 2-year OS and disease-free survival rates of 56% and 22%, respectively (Fig. 5). With 21% of the Evaluabletrimodality group having M1A disease, our survival results are particularly encouraging.

Figure 5.

Overall survival is shown for patients who were evaluable for pathologic response (Evaluabletrimodality) (n = 43) by M classification.

Given the good local and regional control with the trimodality approach presented hereincontrol of systemic disease should be the next focus of development. This regimen should be considered a good platform to evaluate the benefit of the addition of targeted agents, which may have further activity on systemic disease. Future trials should evaluate the addition of targeted therapy using different strategies. One strategy would aim to achieve higher pCR rates with combined targeted therapy and trimodality therapy. We also need to be cognizant that an impact on systemic recurrences with improved systemic agents may not be appreciated by pCR rates. Another design could aim to attack the minimal residual or subclinical disease present after trimodality therapy, with treatment given in an adjuvant setting. Although improving treatment efficacy is the goal, it is likely that patients' tolerance of the additional therapy will be paramount to the success of any of these strategies.


In LAEC, induction therapy with irinotecan plus cisplatin and concurrent radiotherapy is feasible, yields a 72% improvement in dysphagia, and results in a 98% R0 resection rate and a 16% pCR rate. This trimodality approach in resectable patients led to an encouraging median survival of 30 months. It was associated with a significant but manageable toxicity profile as well as transient drops in patient-reported QoL. This regimen clearly warrants further study in select patients.


The authors made no disclosures.