Failure patterns correlate with the proportion of residual carcinoma after preoperative chemoradiotherapy for carcinoma of the esophagus

Authors


Abstract

BACKGROUND

The current study was conducted to test the hypothesis that patterns of failure are correlated with the degree of residual carcinoma after preoperative chemoradiotherapy (CRT) in patients with esophageal carcinoma.

METHODS

The authors analyzed the clinical characteristics of patients with carcinoma of the esophagus who underwent preoperative CRT. The residual carcinoma in the resected specimen was categorized into 3 groups (0%, 1–50%, and > 50%). The initial patterns of failure were analyzed according to these categories.

RESULTS

Of the 235 patients who underwent CRT, 69 (29%) achieved a pathologic complete response (pathCR; Group A), 109 patients (46%) achieved a response but it was less than a pathCR (1–50% residual carcinoma; Group B), and 57 (24%) had no response (> 50% residual carcinoma; Group C). The time to locoregional recurrence was significantly longer for Group A compared with Group C (P = 0.05). The rate of distant metastases was significantly lower in Groups A and B compared with Group C (14% in Group A, 29% in Group B, and 33% in Group C; P = 0.03). The distant metastases-free survival was found to be significantly longer in Groups A and B compared with Group C (Group A vs. Group B, P = 0.01; Group A vs. Group C, P < 0.0001; and Group B vs. Group C, P = 0.03). A significantly higher proportion of patients in the responding groups (Groups A and B) had no disease recurrence compared with Group C (81% in Group A, 67% in Group B, and 61% in Group C; P = 0.04). The overall survival and disease-free survival were found to be significantly longer in Groups A and B compared with Group C.

CONCLUSIONS

Data from the current study demonstrate that the proportion of residual carcinoma after preoperative CRT is significantly correlated with patterns of locoregional and distant failure. Future investigations should focus on reducing the proportion of residual carcinoma and metastatic disease progression in patients with esophageal carcinoma. Cancer 2005. © 2005 American Cancer Society.

Although esophageal carcinoma is rare in the U.S., it is associated with high mortality because of its aggressive nature and poor prognosis.1–7 The overall survival (OS) of patients undergoing primary surgery is predicted based on the pathologic stage and for patients who receive preoperative chemoradiotherapy (CRT), OS is best predicted by posttherapy pathologic stage.8 In the latter category, for example, the survival of patients who achieve a complete pathologic response (pathCR) (defined as the absence of tumor cells in both the esophagus and lymph nodes [T0N0] on microscopic examination of the final surgical pathology specimen) has been correlated with a better outcome than the survival of those patients whose specimens exhibit residual carcinoma.9–16 Although esophageal carcinoma is a systemic disease and local disease, if uncontrolled, does not cause death, patients often die of distant metastases. To our knowledge, these patterns of disease recurrence after preoperative CRT have not been well described to date. Clearly, these patterns are governed by the biology of the tumor but the question remains whether one can identify an intermediate surrogate of the patterns of disease recurrence (e.g., the type of pathologic response observed after preoperative CRT).

The surgical pathology of resected specimens can be divided into 3 groups according to the degree of pathologic response: pathCR (0% residual carcinoma), less than a pathCR but demonstrating some response (1–50% residual carcinoma), and no response (> 50% residual carcinoma). Because the OS among the three groups is distinctly different, we hypothesized that a higher pathologic response would correspond with lower locoregional and distant recurrence rates. A cohort of 235 consecutive patients seen at the University of Texas M. D. Anderson Cancer Center who had undergone preoperative CRT was analyzed.

MATERIALS AND METHODS

Patient Eligibility

Patients with localized, potentially resectable carcinoma of the esophagus who were able to undergo surgery were eligible. All patients were jointly evaluated by a medical oncologist, gastroenterologist, radiation oncologist, and thoracic oncology surgeon prior to therapy. Patients with T1N1 disease, T2N0 or T2N1 disease, T3N0 or T3N1 disease, and, for lower esophageal carcinomas, T1-3 plus celiac lymph node enlargement (N2 or M1a) were eligible. Patients who had T1N0, T4, or M1b carcinoma were excluded. Other exclusions included patients who had uncontrolled medical conditions (i.e., diabetes, hypertension, heart conditions classified as New York Heart Association class III or IV, or psychiatric illness) or who demonstrated a lack of comprehension of the purpose of this clinical investigation or an inability to comply with its requirements. Nutritional counseling was available to a select group of patients based on need and all patients who were participating in the protocol provided written informed consent.

Pretreatment Evaluation

All clinical staging was performed according to the standards of TNM staging as specified by the 6th edition of the American Joint Committee on Cancer (AJCC) system for esophageal carcinoma.17 Clinical staging prior to therapy included a complete history and physical examination, chest radiography, computed tomography (CT) scan of the chest and abdomen, upper gastrointestinal double-contrast barium radiographs, positron emission tomography (PET) (when available), an esophagogastroduodenoscopy with endoscopic ultrasonography (EUS), electrocardiogram, sequential multiple analysis–12 channel biochemical profile (SMA-12), electrolytes, complete blood count including platelet count, and serum baseline carcinoembryonic antigen level. Additional preoperative screening studies were performed as needed.

Evaluation during Therapy

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

Chemoradiotherapy

Patients entering various approved protocols at the University of Texas M. D. Anderson Cancer Center were included in this analysis. Patients treated with preoperative CRT off protocol also were included in this analysis. The analysis was approved by the Institutional Review Board. Some patients received induction chemotherapy prior to CRT whereas others received preoperative CRT only. The radiation dose was either 45 grays (Gy) in 25 fractions or 50.4 Gy in 28 fractions. Patients underwent simulation by standard methods using esophagoscopy, esophagography, and CT information. Using megavoltage (≥ 6 megavolts) equipment and CT information to help in the construction of the treatment portals, a two-field (anterioposterior/posterioanterior fields) to four-field (anterioposterior/posterioanterior fields) technique was applied. The superior and inferior borders of the field extended from 5 cm beyond the edges of the macroscopic tumor volume, with lateral borders at 2 cm beyond it. Radiotherapy was prescribed to cover at least 95% of this field plus a 5-mm margin.

The chemotherapy agents used during either induction or concurrently with radiotherapy predominantly included cisplatin, 5-fluorouracil, irinotecan, and paclitaxel or docetaxel.

Evaluation Prior to Surgery

Approximately 4–6 weeks after the completion of CRT, patients underwent a complete staging workup including a PET scan (when available) but not EUS.

Surgery

A curative radical en bloc esophagectomy was attempted to resect the specimen for posttherapy pathologic staging according to the AJCC staging system. Patients received one of three approaches: transthoracic surgery, transhiatal surgery, or a threefield (bilateral cervical, mediastinal, and abdominal) lymphadenectomy.

Postsurgical Pathologic Evaluation

Postsurgical staging was performed pathologically based on residual carcinoma status as determined in the resected specimen. Areas with ulceration or scarring indicating the therapy field were submitted completely for histologic examination to confirm the macroscopically identified absence of residual carcinoma. A pathologist (T.T.W.) who was blinded to the type of therapy or outcome reviewed all hematoxylin and eosin-stained sections, including surgical margins and lymph nodes. The extent of residual carcinoma in the specimen was assessed semiquantitatively based on the estimated percentage of residual carcinoma in relation to the total carcinoma area, including the amount of radiation-induced tissue injury, in mural histologic sections. Patients were assigned to one of three groups based on the extent of residual carcinoma. Group A was comprised of patients who achieved a pathCR, Group B was comprised of patients who achieved some response (defined as 1–50% residual carcinoma in the resected specimen), and Group C was comprised of patients with no response (defined as greater than 50% residual carcinoma in the resected specimen). A pretreatment classification of M0 was a requirement for classification as having achieved a pathCR.

Follow-Up

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

Statistical Analysis

The Pearson chi-square test or the Fisher exact test were used to compare the distribution of categoric parameters between combinations of two groups of the three total groups (Group A [pathCR], Group B [1–50% residual carcinoma], and Group C [> 50% residual carcinoma]). Analysis of variance was used to compare the distribution of parameters among all three groups.

Kaplan–Meier survival curves were plotted for OS, disease-free survival (DFS), locoregional recurrence-free survival, and distant metastases-free survival. The log-rank test was used to evaluate the statistical significance of the differences noted between combinations of two of the three total groups.

Statistical analysis was performed using SPSS software (version 11.5.2.1 for Windows; SPSS Inc., Chicago, IL). All statistical analyses were performed with a two-sided significance level of 0.05.

RESULTS

Patient Characteristics

In the current study, a total of 235 consecutive patients who were evaluated at the University of Texas M. D. Anderson Cancer Center between 1985–2003, were histologically confirmed to have either adenocarcinoma or squamous cell carcinoma of the esophagus, and who underwent pretreatment clinical staging followed by preoperative CRT were analyzed. Among these patients, there were 13 patients with localized Stage IV disease (M1a) who also underwent surgery and were included in the current analysis. Three patients with borderline T4 tumors also were included in this analysis. Of the 235 patients, 69 (29%) were in Group A, 109 (46%) were in Group B, and 57 (24%) were in Group C. Characteristics of the patients and results from the statistical analyses comparing each group are summarized in Table 1.

Table 1. Characteristics and Correlations among Various Response Groups
 Group Aa (n = 69)Group Ba (n = 109)Group Ca (n = 57)Total (n = 235)P1bP2bP3bP4b
No.(%)No.(%)No.(%)No.(%)
  • GEJ: gastroesophageal junction; AJCC: American Joint Committee on Cancer; chemo: chemotherapy; RT: radiotherapy; Gy: grays.

  • a

    Group A was comprised of 69 patients who achieved a pathologic complete response (pathCR), Group B was comprised of 109 patients who achieved a response but it was less than a pathCR (1–50% residual carcinoma), and Group C was comprised of 57 patients who had no response (> 50% residual carcinoma).

  • b

    P1: overall significance comparison; P2: significance Group A vs. Group B; P3: significance Group A vs. Group C; P4: significance Group B vs. Group C.

  • c

    Thoracic esophagus only.

  • d

    M1a. The median follow-up time was 37 months.

Gender            
 Male57(83)96(88)50(88)203(86)0.550.310.430.95
 Female12(17)13(12)7(12)32(14)    
Tumor locationc            
 Upper/middle10(15)11(10)9(16)30(13)0.510.380.840.28
 Lower/GEJ59(86)98(90)48(84)205(87)    
Histology 1            
 Adenocarcinoma56(81)94(86)43(75)193(82)0.220.370.440.08
 Squamous cell13(19)15(14)14(25)42(18)    
T classification        0.540.770.300.49
 T223(33)31(28)16(28)70(30)    
 T346(67)77(71)39(68)162(69)    
 T40(0)1(1)2(4)3(1)    
N classification        0.420.270.240.79
 N040(58)54(50)27(47)121(52)    
 N129(42)55(51)30(53)114(49)    
Clinical M           0.37
 M069(100)102(94)51(90)222(95)    
 M1d0(0)7(6)6(11)13(6)    
Clinical AJCC stage        0.070.060.020.55
 Stage II45(65)60(55)32(56)137(58)    
 Stage III24(35)42(39)19(33)85(36)    
 Stage IVζ0(0)7(6)6(11)13(6)    
Chemoradiation sequence        0.030.260.010.06
 Chemo/RT32(46)60(55)40(70)132(56)    
 Chemo → Chemo/RT37(54)49(45)17(30)103(44)    
Preoperative radiation dose        0.910.970.700.70
 ≤ 45 Gy49(71)75(71)38(68)162(70)    
 > 45 Gy20(29)31(29)18(32)69(30)    

Overall, a majority of the cohort were male (n = 203; 86%). The location of the primary tumor was the distal esophagus or the gastroesophageal junction for the majority of patients (n = 205; 87%), and most patients had adenocarcinoma (n = 193; 82%). One hundred thirty-seven patients (58%) had pretreatment clinical Stage II carcinoma, 85 (36%) had Stage III carcinoma, and 13 (5%) had Stage IV carcinoma (Table 1).

Patterns of Failure

The median follow-up time was 37 months. Patterns of failure for patients in each group are summarized in Table 2. Comparative histograms of the percentage of patients in each group with specific patterns of failure are shown in Figure 1. The rate of locoregional recurrence was not found to be significantly different among the 3 groups at the stated median follow-up time of 37 months. However, there was a significantly longer time to locoregional recurrence noted in Group A compared with Group C (median time to locoregional recurrence was not reached for both groups; P = 0.05). Ten patients (14%) in Group A, 32 patients (29%) in Group B, and 19 patients (33%) in Group C had distant metastases (P = 0.03). The median time to distant metastases was not reached for Group A and was 34 months (range, 0 –> 100 months) for Group B and 12 months (range, 0–> 39 months) for Group C (P = 0.0002). The time to distant metastases was significantly longer for Group A compared with Group B (P = 0.01), Group A compared with Group C (P < 0.0001), and Group B compared with Group C (P = 0.03). Distant metastases-free survival curves are shown in Figure 2.

Table 2. Comparison of Patterns of Failure among Various Pathologic Response Groupsa
 Group AbGroup BbGroup CbP valuec
No.(%)No.(%)No.(%)
  • a

    Comparison of failure at a median follow-up time of 37 months.

  • b

    Group A was comprised of 69 patients who achieved a pathologic complete response (pathCR), Group B was comprised of 109 patients who achieved a response but it was less than a pathCR (1–50% residual carcinoma), and Group C was comprised of 57 patients who had no response (> 50% residual carcinoma).

  • c

    Analysis of variance was performed comparing percentages from the three groups.

Total69 109 57  
Locoregional recurrence5(7)12(11)8(14)0.47
Distant metastases10(14)32(29)19(33)0.03
Any recurrence13(19)36(33)22(39)0.04
Figure 1.

Histogram for the percentage of each pathologic response group with specified failure and analysis of variance P values. Analysis was performed after a median follow-up of 37 months.

Figure 2.

Kaplan–Meier plot for distant metastases-free survival according to pathologic response. Group A versus Group B, P = 0.01; Group A versus Group C, P < 0.0001; and Group B versus Group C, P = 0.03.

Disease-Free Survival

Fifty-six patients (81%) in Group A, 73 patients (67%%) in Group B, and 35 patients (61%) in Group C had no failure after resection (P = 0.04). The median DFS time was not reached (range, 3–> 69 months) for Group A, but was 20 months (range, 0–> 100 months) for Group B and 12 months (range, 0–> 39 months) for Group C (P = 0.0001). The DFS was significantly longer for Group A compared with Group B (P = 0.02), Group A compared with Group C (P < 0.0001), and Group B compared with Group C (P = 0.01). The DFS analysis is summarized in Table 2 and Figure 3.

Figure 3.

Kaplan–Meier plot for disease-free survival according to pathologic response. Group A versus Group B, P = 0.02; Group A versus Group C, P < 0.0001; and Group B versus Group C, P = 0.01.

Overall Survival

Fifty-one patients (74%) in Group A, 71 patients (65%) in Group B, and 23 patients (40%) in Group C were alive at the time of last follow-up (P < 0.001). The median OS time of the patients in Group A was 133 months (range, 1–133 months, according to last follow-up), whereas that for Group B was 42 ± 6 months (range, < 1–> 186 months), and the median OS time for Group C was 14 ± 1 months (range, 1–> 178 months) (P < 0.0001). The OS was found to be significantly longer for Group A compared with Group C (P < 0.0001) and for Group B compared with Group C (P = 0.0001).

Impact of Therapy Sequence and Radiation Dose

Although induction chemotherapy was found to be significantly correlated with greater pathologic response when analyzing only groups A and B, and when analyzing only groups B and C, these results are preliminary. Radiation dose was not found to have any impact on the type of pathologic response achieved (Table 1).

DISCUSSION

Clinical staging of carcinoma of the esophagus is of paramount importance in the determination of appropriate initial therapy. However, the patterns of failure and survival of the patients cannot be reliably determined based on the clinical stage of disease. Therefore, several patients with the same clinical disease stage who are treated similarly often have a somewhat heterogeneous outcome. The reason most likely is due to inherent and significant heterogeneity in the clinical biology of tumors.

The results of the current study demonstrate that if patients with localized esophageal carcinoma are treated similarly (e.g., with preoperative CRT), the pathologic responses vary based on the inherent chemoradioresistance of the tumor. These differing patterns of pathologic response appear to be correlated directly with the patterns of disease recurrence, DFS, and OS. There are two important aspects to be considered. First, are we simply uncovering the true clinical biology of these tumors by treating them this way (i.e., by observing the types of pathologic response after preoperative CRT) or are we actually altering the clinical biology? This type of question would require not only well designed prospective randomized trials but also the extensive evaluation of the clinical biology of the tumor. Second, by observing various degrees of pathologic response and correlations with the patterns of disease recurrence, DFS, and OS, one could consider the degree of pathologic response to be a useful surrogate marker for eventual patient outcome. If one accepts that the type of pathologic response is a surrogate marker for patient outcome, it could shorten the follow-up duration required in Phase III studies (it could be an intermediate for the typical endpoints such as DFS or OS).

The current study data also demonstrate that esophageal carcinomas that are locally chemoradioresistant are also by nature metastatic. Therefore, only 14% of the patients in Group A had distant metastases whereas 29% of the patients in Group B and 33% of the patients in Group C had distant metastases (P = 0.03). The median time to distant metastases was not reached for Group A but was found to be 34 months for Group B and only 12 months for Group C (P = 0.0002). In addition, significant differences were noted with regard to metastasis-free intervals. Time to distant metastases was found to be significantly longer for Group A compared with Group B (P = 0.01), Group A compared with Group C (P < 0.0001), and Group B compared with Group C (P = 0.03). This simply means that patients cannot achieve a better outcome if treatment strategies predominantly focus on local control. One must effectively treat occult metastatic disease as well. The question remains as to how one should do that. One assumption is that the primary tumor contains the targets that are also present in the metastases. Therefore, if one can discover targets in the primary tumor for which to develop systemic strategies, it may be of substantial value. However, this notion remains hypothetical at present.

For patients undergoing surgery as their primary therapy, locoregional recurrence rates have been reported to be unacceptably high after an R0 resection of carcinoma of the esophagus (greater than 50%).18 Conversely, with a similar median follow-up of 37 months and a similar patient population (there were no patients with Stage I disease in the current series), the data from the current study suggest that preoperative therapies may have effectively reduced the rate of locoregional disease recurrence (reported to be less than 15% in all 3 pathologic response groups). The low rate of disease recurrence may reflect the use of induction chemotherapy, more active chemotherapy combinations, effective radiation sensitizers, or better surgical and radiation techniques. The rate of locoregional recurrence was not found to be significantly different among our three groups. Nevertheless, clinical biology again expressed itself in that locoregional disease recurrence occurred more rapidly in Group C patients compared with Group A patients (P = 0.05).

The results of the current analysis suggest that the proportion of residual carcinoma in the resected specimen after preoperative CRT is a reflection of the clinical biology of the tumor and can serve as a surrogate marker for future Phase III trials because it is correlated directly with the patterns of failure, DFS, and OS. The resistant localized esophageal carcinoma also has high metastatic potential and further research should focus on the discovery of specific targets to treat localized disease as well as to reduce metastatic progression. We propose that an understanding of molecular biology and a patient's genetics may be able to predict the clinical biology of esophageal carcinoma.

Ancillary