Histologic subtypes as determinants of outcome in esophageal carcinoma patients with pathologic complete response after preoperative chemoradiotherapy




The current study tested the hypothesis that the clinical outcome of patients with localized esophageal carcinoma after preoperative chemoradiotherapy (CTRT) depends on histology.


The authors stratified patients by adenocarcinoma (ACA) or squamous cell carcinoma (SCC) and compared the overall survival (OS) and patterns of failure among patients achieving pathologic complete response (pathCR) and < pathCR after preoperative CTRT. A correlation between baseline clinical stage and posttherapy pathologic response was made for ACA and SCC.


Of the 235 patients who underwent preoperative CTRT, 42 (18%) had SCC and 193 (82%) had ACA. In the ACA group, 56 patients (29%) achieved a pathCR and in the SCC group 13 patients (31%) achieved a pathCR. In the ACA group, a larger proportion of pathCR patients (n = 44; 79%) than < pathCR patients (n = 82; 60%) were alive at the time of last follow-up (P = 0.01) and pathCR patients had a longer OS than < pathCR patients (P = 0.0006). However, in the SCC group OS or proportion alive did not differ significantly between pathCR and < pathCR patients (P ≥ 0.05). In the ACA group, a greater portion of < pathCR patients (32%) than pathCR patients (16%) had distant metastases (P = 0.02) and the distant metastases-free survival of pathCR patients was longer than that of < pathCR patients (P = 0.0012). In the SCC group, the proportion or time to distant-metastases did not differ significantly. Pretreatment clinical stage did not correlate with pathologic response for either histology.


The results of the current study suggest that the clinical biology of SCC and ACA is different after CTRT. An investigation of molecular and patient genetics is needed to improve therapy. Cancer 2006. © 2005 American Cancer Society.

Esophageal carcinoma is rare in the U.S. but is much more prevalent in other countries and carries a dismal prognosis.1–7 Of all the treatments available, chemoradiation therapy (CTRT) followed by resection has been the adopted strategy by many centers when dealing with localized carcinoma of the esophagus; often, no distinction is made with regard to the histology of the tumor in recommending this strategy.8–12 However, it may be important to consider the clinical biology based on histology in greater detail to derive clues to improve therapy and to individualize therapy. To begin with, a review of the impact of treatment on patient outcome by histologic subtype, either adenocarcinoma (ACA) or squamous cell carcinoma (SCC), could be intriguing because the lifestyle factors that are associated with these common histologies are different; therefore, the survival pathways and genetics of these two malignancies also might be different.

After a curative surgical resection, patients with SCC are reported to survive significantly shorter than those with ACA.13 The reason for such outcome differences after surgery are only speculative but may be related to differences in molecular biology and resultant clinical biology. It is known that a correlation exists between pathologic complete response (pathCR) and overall survival (OS) in patients treated with preoperative CTRT.14 However, whether this clinical behavior holds true for both ACA and SCC histologic subtypes is not known. Similarly, it appears that the patterns of failure (local, regional, distant, or death) correlate well with the degree of pathologic response observed.15 This has also not been defined based on histologic subtypes. Thus, we hypothesized that the patterns of failure and OS might be different in ACA and SCC patients who achieve a pathCR after preoperative CTRT.

In the current study, we retrospectively compared OS and patterns of failure between pathCR patients and also < pathCR patients within each of the two histologies. A total of 235 consecutive patients who had undergone preoperative CTRT at the University of Texas M. D. Anderson Cancer Center were analyzed.


Patient Eligibility

Operable patients with localized, potentially resectable carcinoma of the esophagus who were evaluated at the University of Texas M. D. Anderson Cancer Center from 1985–2003 were eligible for this retrospective analysis. Before therapy, each patient was evaluated by a medical oncologist, gastroenterologist, radiation oncologist, and thoracic oncology surgeon. Patients with T2 or T3, N0 or N1 (clinical TNM Stage II or III) esophageal carcinoma were included (also included were patients in whom the gastroesophageal junction was involved with M1a lymph nodes), whereas patients with T1N0, T4, or M1b disease were excluded, as were patients who had uncontrolled medical conditions (e.g., diabetes mellitus, hypertension, heart conditions classified as New York Heart Association Class III or IV, or psychiatric illnesses), were unable to comprehend the purpose of this clinical investigation, or were unable to comply with its requirements. Nutritional counseling was available to a select group of patients based on need and all protocol-participating patients signed a written informed consent form. This analysis was approved by the Institutional Review Board.

Pretreatment Evaluation

All clinical staging was performed according to American Joint Committee on Cancer (AJCC) criteria for staging esophageal carcinoma.16 Clinical staging was performed before therapy and included a complete history and physical examination; chest radiography; computed tomography (CT) scans of the chest and abdomen; upper gastrointestinal double-contrast barium radiography; positron emission tomography (when available); an esophagogastroduodenoscopy with or without endoscopic ultrasonography (EUS); electrocardiography; SMA-12; and a complete blood count including platelet count and measurement of serum, electrolyte, and baseline carcinoembryonic antigen levels. Additional preoperative screening studies were performed as needed.

Evaluation During Therapy

Patient symptoms and the results of blood tests were closely monitored during therapy. Patients underwent repeat endoscopy or other imaging studies as warranted by their clinical symptoms.

Chemoradiation Therapy

Some patients received induction chemotherapy before chemoradiotherapy and others received preoperative chemoradiotherapy only. The radiation dose was either 45 grays (Gy) in 25 fractions or 50.4 Gy in 28 fractions. Standard methods (i.e., esophagoscopy, esophagography, and CT) were used for simulation. Megavoltage (6-megavolt) equipment was used and a 2-to-4 field technique was applied. The superior and inferior borders of the field extended from 5 cm beyond the cephalad and caudal margins of the carcinoma, with lateral borders at 2 cm beyond the edges. Radiotherapy was prescribed to cover at least 95% of this field plus a 5-mm margin.

Chemotherapy agents used either during induction or concurrently with radiotherapy predominantly included cisplatin, 5-fluorouracil, camptothecin-11, and paclitaxel or docetaxel.

Evaluation Before Surgery

Approximately 4–6 weeks after the completion of CTRT, patients underwent another complete staging workup including positron emission tomography (when available) but not EUS.


A curative radical en bloc esophagectomy was attempted in all patients, and the resected specimens analyzed to determine the posttherapy pathologic AJCC stage. One of three approaches were used in patients: transthoracic surgery, transhiatal surgery, or three-field lymphadenectomy.

Postsurgical Evaluation

Postsurgical staging was based on pathologic findings in the resected specimen, in particular the residual carcinoma status. Areas demonstrating ulcer or scar formation, indicating the therapy field, were submitted intact for histologic examination to confirm that the specimen was from a macroscopically identified absence of residual carcinoma. A pathologist (T.T.W.), who was blinded to the type and outcome of therapy, reviewed all hematoxylin and eosin-stained sections, including margins and lymph nodes. Each specimen was categorized in one of two groups: pathCR or < pathCR.


After surgery, patients were followed every 3 months for the first year and then every 6 months for the next 2 years, and finally every year for 2 additional years. Patients underwent CT scans every 6 months during the first 3 years. Endoscopy with biopsy was alternated with upper gastrointestinal radiographs at each follow-up.

Statistical Methods

A chi-square test or Fisher exact test, when necessary, were performed to compare survival and failure pattern rates between pathCR and < pathCR patients. Multivariate Cox regression analysis was performed when adjusting for possibly confounding variables. Kaplan–Meier curves were plotted for OS, disease-free survival (DFS), locoregional recurrence-free survival, and distant metastases-free survival for pathCR and < pathCR patients by histology. A log-rank test was used to compare survival curves of pathCR and < pathCR patients.

Statistical analysis was performed using SPSS software (version for Windows; SPSS Inc., Chicago, IL). All statistical analyses were performed with a two-sided significance value of 0.05.


Patient Characteristics

Two hundred thirty-five consecutive patients evaluated at the University of Texas M. D. Anderson Cancer Center between 1985–2003 and histologically confirmed to have either ACA or SCC of the esophagus who underwent pretreatment clinical staging followed by preoperative chemoradiotherapy were analyzed. Of these, 193 patients (82%) were diagnosed with ACA and 42 (18%) with SCC. Of the ACA patients, 56 (29%) achieved a pathCR whereas of the SCC patients, 13 (31%) achieved a pathCR.

Survival Analyses

Of the ACA patients, a significantly larger proportion of pathCR patients (n = 44; 79%) than < pathCR patients (n = 82; 60%) were alive at the time of last follow-up after adjusting for clinical stage of disease, gender, chemoradiation sequence, and age (P = 0.002) at a median follow-up time of 37 months. Among SCC patients, the proportion of pathCR patients (n = 7; 54%) and < pathCR patients (n = 12; 41%) that remained alive at the time of last follow-up did not differ significantly after adjusting for clinical stage of disease, gender, chemoradiation sequence, and age (P = 0.6). Data are summarized in Table 1.

Table 1. Proportion of Patients with Failure, Stratified by Pathologic Response
 ACA (n = 193)SCC (n = 42)
pathCR (n = 56)< pathCR (n = 137)pathCR (n = 13)< pathCR (n = 29)
  • ACA: adenocarcinoma; SCC: squamous cell carcinoma; pathCR: pathologic complete response; < pathCR: less than a pathologic complete response.

  • a

    “Event” indicates death or disease recurrence.

  • Numbers in bold indicate a significant difference between the proportion of patients with a pathCR and those with a < pathCR after adjusting for clinical stage of disease, gender, chemoradiation sequence, and age.

Distant metastases916443218724
Locoregional recurrence59151000517

Among ACA patients, pathCR patients survived significantly longer (median OS of 133 mos) than < pathCR patients (median OS of 32 ± 5 mos; P = 0.0006). Among SCC patients, the OS of pathCR patients (median OS of 29 mos) and that of < pathCR patients (median OS of 35 ± 16 mos) did not vary significantly (P = 0.6) (Fig. 1).

Figure 1.

Overall survival analysis using the Kaplan–Meier method, stratified by the pathologic response of (A) adenocarcinoma patients and (B) squamous cell carcinoma patients. P-values were determined using the log-rank test. pathCR: pathologic complete response; < pathCR: no pathologic complete response

Among the ACA patients, a significantly larger portion of pathCR patients (n = 40; 71%) than < pathCR patients (n = 70; 51%) survived without evident malignancy after adjusting for clinical stage of disease, gender, chemoradiation sequence, and age (P = 0.002). Among SCC patients, the proportion of pathCR patients (n = 6; 46%) and < pathCR patients (n = 11; 38%) did not differ significantly after adjusting for clinical stage of disease, gender, chemoradiation sequence, and age (P = 0.5) (Table 2).

Table 2. Distribution of Clinical Stage of Disease, Stratified by Pathologic Responsea
 ACA (n = 193)SCC (n = 42)
pathCR (n = 56)< pathCR (n = 137)pathCR (n = 13)< pathCR (n = 29)
  • ACA: adenocarcinoma; SCC: squamous cell carcinoma; pathCR: pathologic complete response; < pathCR: no pathologic complete response.

  • a

    There was no significant correlation noted between pathologic response and clinical stage within either tumor histology.

Stage II3563705610772279
Stage III21385544323621

DFS curves followed patterns similar to OS curves during the 140 months analyzed for ACA and 200 months analyzed for SCC patients.

Multivariate Analyses

We performed multivariate analyses in three cohorts: ACA patients, SCC patients, and all patients combined. The factors considered included clinical stage of disease, gender, use of chemotherapy and chemoradiation versus chemoradiation, pathCR versus < pathCR, and age. Only pathologic response was found to be an independent predictor in ACA patients (P = 0.002) and the entire cohort (P = 0.007) but not for SCC patients (P = 0.56) (Table 3).

Table 3. Multivariate Analysis by Histologic Subtypes and the Entire Population
 FrequencyEvent: death95% CI for HREvent: recurrence or death95% CI for HREvent: distant recurrence95% CI for HREvent: local or regional recurrence95% CI for HR
P valueHRLowerUpperP valueHRLowerUpperP valueHRLowerUpperP valueHRLowerUpper
  1. ACA: adenocarcinoma; HR: hazards ratio; 95% CI: 95% confidence interval; Chemo: chemotherapy; XRT: radiation therapy; pathCR: pathologic complete response; non-pathCR: no pathologic complete response; SCC: squamous cell carcinoma.

Multivariable Cox regression analysis–ACA (n = 193)
Clinical stage-esophagus 0.476   0.578   0.193   0.494   
 Stage II (reference)105 1.000   1.000   1.000   1.000  
 Stage III760.7780.9250.5381.5910.9121.0270.6391.6500.3871.3120.7092.4300.8690.9160.3242.590
 Stage IVA120.2841.6310.6663.9920.3011.5400.6803.4900.0732.3770.9246.1200.2752.4100.49611.702
 Female (reference)13 1.000   1.000   1.000   1.000  
Chemoradiation sequence                 
 Chemo/XRT (reference)104 1.000   1.000   1.000   1.000  
 Chemo → Chemo/XRT890.7800.9310.5621.5410.6170.8920.5701.3960.7860.9240.5221.6350.1830.5140.1931.368
pathCR vs. < pathCR                 
 pathCR (reference)56 1.000   1.000   1.000   1.000  
 < pathCR1370.0022.9991.5115.9530.0022.5061.3954.5000.0092.7231.2825.7830.2411.9370.6415.852
Multivariable Cox regression analysis–SCC (n = 42)
Clinical stage-esophagus 0.814   0.846   DID NOT CONVERGE DID NOT CONVERGE 
 Stage II (reference)32 1.000   1.000          
 Stage III90.6880.7490.1823.0780.8180.8620.2443.050        
 Stage IVA10.7331.5770.11521.6670.6681.7390.13921.816        
 Female (reference)19 1.000   1.000          
Chemoradiation sequence                 
 Chemo/XRT (reference)28 1.000   1.000          
 Chemo → Chemo/XRT140.8841.1070.2824.3500.9051.0770.3193.641        
pathCR vs. < pathCR                 
 pathCR (Reference)13 1.000   1.000          
 < pathCR290.5651.3780.4634.1010.5261.3790.5113.726        
Age 0.2631.0360.9741.1010.1201.0470.9881.110        
Multivariate Cox regression analysis-ALL (n = 235)
Clinical stage-esophagus 0.353   0.441   0.117   0.697   
 Stage II (reference)137 1.000   1.000   1.000   1.000  
 Stage III850.8470.9530.5861.5510.8161.0530.6831.6220.4561.2490.6972.2390.9321.0400.4222.566
 Stage IVA130.1801.7600.7704.0190.2021.6450.7663.5310.0382.5201.0516.0420.4011.9360.4159.038
 ACA (reference)193 1.000   1.000   1.000   1.000  
 Female (reference)32 1.000   1.000   1.000   1.000  
Chemoradiation sequence                 
 Chemo/XRT (reference)132 1.000   1.000   1.000   1.000  
 Chemo → Chemo/XRT1030.9851.0040.6371.5820.8890.9710.6451.4620.6360.8770.5091.5120.4640.7210.3001.730
pathCR vs. < pathCR                 
 pathCR (Reference)69 1.000   1.000   1.000   1.000  
 < pathCR1660.0072.0941.2263.5770.0061.9341.2033.1110.0052.7321.3495.5310.1532.0930.7595.772

Patterns of Failure

Of the ACA patients, a significantly greater portion of < pathCR patients (n = 44; 32%) than pathCR patients (n = 9; 16%) had distant metastases after adjusting for clinical stage of disease, gender, chemoradiation sequence, and age (P = 0.009). Among SCC patients, the proportion of pathCR patients (n = 1; 8%) and < pathCR patients (n = 7; 24%) with distant metastases did not differ significantly after adjusting for clinical stage of disease, gender, chemoradiation sequence, and age (P = 0.4) (Table 1). Among ACA patients, the distant metastases-free survival of pathCR patients was significantly longer (median OS was not reached) than that of < pathCR patients (median OS of 17 ± 2 mos; P = 0.0012). Among the SCC patients, the OS of pathCR patients and that of < pathCR patients did not vary significantly and the median of both groups was not reached (P = 0.3) (Fig. 2).

Figure 2.

Distant metastases analysis using the Kaplan–Meier method stratified by the pathologic response of (A) adenocarcinoma patients and (b) squamous cell carcinoma patients. P-values were determinted using the log-rank test. pathCR: pathologic complete response; < pathCR: no pathologic complete response

The proportion of patients with locoregional recurrence after adjusting for clinical stage of disease, gender, chemoradiation sequence, and age or time to locoregional recurrence did not differ significantly within either of the two histologies (data not shown).

Clinical Stage Distribution before Therapy

Baseline clinical stage did not appear to be correlated significantly with pathologic response within ACA patients (P = 0.4) or SCC patients (P = 1.0) (Table 2).


The outcome of patients with localized carcinoma of the esophagus remains dismal despite some advances in chemotherapy agents, an improved understanding of chemoradiobiology, and improvement in multidisciplinary care. The pathCR rates are approximately 27%.14 The therapeutic approach has been empiric for decades. Any substantial gains may not be anticipated by continued empiric approaches. Therefore, more attention to clinical biology may be needed and, more important, molecular biology and patient genetics. Deriving therapy based on the molecular biology of the tumor and patient genetics is a major challenge but a much-needed effort. However, at the current time, there is an opportunity to study the clinical biology of esophageal cancer.

ACA and SCC appear to have a distinct clinical profile. ACA occurs predominantly in white men and is often associated with increased body mass index and chronic gastroesophageal reflux disease, whereas SCC, although more common in men, is predominantly a lifestyle-related malignancy (excessive smoking, drinking, and poor nutrition). SCC is frequently present in the upper half of the esophagus and ACA in the lower half. Therefore, these two major histologic subtypes are likely to have different molecular profiles and patient genetics. However, when deciding on therapy, we are not equipped to take advantage of any inherent differences between the two subtypes.

We hypothesized that if ACA and SCC patients with localized tumors were treated with similar strategy (i.e., CTRT followed by surgery), the patient outcome (OS and patterns of failure) would be different. We based this hypothesis on the assumption that ACA and SCC might be employing different survival pathways and therefore would have differing chemoradioresistance profiles as well as differing metastatic potential. We were partially incorrect in that assumption because the pathCR rate was found to be similar for both histologic subtypes (29% in ACA and 31% in SCC). However, the most intriguing findings were observed with the outcome analysis of pathCR and < pathCR patients in both histologic subtypes.

ACA patients achieving a pathCR lived significantly longer than ACA patients with < pathCR; however, there was no advantage noted for SCC patients who achieved a pathCR compared with SCC patients who achieved < pathCR (Fig. 1). The explanation for this intriguing finding was found in the patterns of failure. Among the ACA patients, the occurrence of metastases was significantly less common in pathCR patients compared with < pathCR patients (P = 0.02), but in SCC patients the frequency of metastases for the pathCR and < pathCR patients was similar. This suggests that, even though a higher proportion of Stage II patients (77%) were in the SCC pathCR group compared with the ACC pathCR patients (63%), pathCR SCC patients have higher metastatic potential than pathCR ACC patients. However, among patients with < pathCR, ACC patients appear to fare poorly because they often have lymph node-positive disease more often than SCC patients (data not shown). Therefore, systemic control is more important for pathCR SCC patients and < pathCR ACC patients. Therefore, discovering targets through the molecular biology of cancer would be central to future investigation of these malignancies.

The current study also makes a unified observation regarding the inability of clinical stage of disease to predict patient outcome. No significant correlation between pathologic response and baseline was noted in eeither the SCC or ACA patients. Therefore, the clinical stage of carcinoma of the esophagus also needs further refinement to help individualize therapy. We acknowledge a possible criticism of the current study findings in that different chemotherapy sensitizer combinations could influence the outcome; however, we found no correlation between the type of preoperative chemoradiotherapy and the type of pathologic response achieved and, similarly, preoperative therapy (radiotherapy dose or type of chemotherapy combination) did not correlate with OS or DFS.17

Localized ACA and SCC appear to demonstrate differing clinical biology after preoperative CTRT. Data from the current study suggest that future efforts to individualize therapy for patients with carcinoma of the esophagus should be based on studies of clinical biology, molecular biology of the tumor, and patient genetics. Empiric therapy without regard to these differences would result in minimum, if any, advantage for our patients.