The degree of benefit from palliative chemotherapy differs widely among patients with metastatic esophageal squamous cell carcinoma (MESCC). The purpose of this study was to develop and validate a prognostic nomogram to predict survival and aid physicians and patients in the decision-making process regarding treatment options.
Clinicopathologic variables and treatment outcomes of 239 patients who were diagnosed with MESCC and received either fluorouracil/cisplatin (FP) or capecitabine/cisplatin (XP) as first-line chemotherapy were reviewed. A nomogram was developed as a prognostic scoring system incorporating significant clinical and laboratory variables based on a multivariate Cox proportional hazards regression model. An independent series of 61 MESCC patients treated with FP served as an independent data set for nomogram validation.
No difference in response rate was observed between the FP group (44.8%) and the XP group (54.2%). Similarly, no significant differences in median progression-free survival and median overall survival were observed between regimen groups. Multivariate analysis showed that poor performance status (Eastern Cooperative Oncology Group [ECOG] status≥2), weight loss (10% of the weight loss for 3 months), low albumin level (≤3.5 g/dL), and absence of previous esophagectomy at the time of chemotherapy were significantly associated with low OS in both groups (p<0.05). Based on these findings, patients were classified into favorable (score, 0 to 90), intermediate (91-134), and poor (>135) prognostic groups. The median survival for those with a favorable ECOG was 13.8 months (95% confidence interval [CI], 10.8 to 18.6 months), for intermediate 11.2 months (95% CI, 8.7 to 11.9 months), and for poor, 7.0 months (95% CI, 3.6 to 10.0 months). External validation of the nomogram in a different patient cohort yielded significantly similar findings.
The nomogram described here predicts survival in MESCC patients and could serve as a guide for the use of FP/XP chemotherapy in MESCC patients.
The incidence of esophageal cancer has increased in recent years, reflecting a shift in histologic type and primary tumor location in the West [
Compared to other solid tumors, randomized phase III trials to test the efficacy of cytotoxic chemotherapy are lacking for metastatic esophageal squamous cell carcinoma (MESCC). One randomized phase II study published in Europe more than 20 years ago showed that a combination of cisplatin/fluorouracil (5-FU) (FP) was superior to cisplatin alone (response rate, 35% vs. 19%, respectively), although no clear improvement in overall survival (OS) was observed (33 weeks vs. 28 weeks, respectively) [
The combination of cisplatin and 5-FU is the most commonly used regimen as first-line chemotherapy for MESCC. However, the differences in treatment responses and toxicity levels in these two different clinical trial study populations highlight the difficulty in determining the most effective treatment regimens for MESCC patients. These studies further illustrate the importance of identifying patients who would benefit from cytotoxic chemotherapy. Here, we developed and validated a prognostic nomogram to predict survival that will aid physicians and patients in making an appropriate clinical decision regarding treatment.
This study was approved by the Institutional Review Board (IRB) of the Samsung Medical Center, Seoul, South Korea and is registered on clinicaltrials.gov (NCT#NCT 01472419). The informed consent form was waived by the IRB. Between January 2000 to December 2010, 239 patients were diagnosed with recurrent MESCC and received either FP or capecitabine/cisplatin (XP) as first-line chemotherapy. The following clinicopathologic variables and treatment outcomes were collected: patient demographics, laboratory data at the time of first-line palliative chemotherapy, surgical record, pathologic report, TNM stage, treatment record, and vital status.
FP and XP regimens used to treat the patients are described below. In the FP group, 5-FU and cisplatin were administered in 4-day courses, once every 21 days. 5-FU (1,000 mg/m2) was infused over 24 hours during the four days. Cisplatin was given in one dose of 60 mg/m2 at the beginning of each cycle. In the XP group, capecitabine (1,250 mg/m2 bid) was administered for 14 days and cisplatin was given in one dose of 60 mg/m2 at the beginning of each cycle. The clinical tumor response was assessed according to World Health Organization (WHO) criteria or Response Evaluation Criteria in Solid Tumors criteria (RECIST 1.0).
As an external validation data set, 61 patients with MESCC who received palliative fluoropyrimidine/platinum chemotherapy (XP or FP) at Northern France Cancer Center were included.
The primary endpoints of the study were OS and progression-free survival (PFS). OS was measured from the date of first chemotherapy to the date of death or the last follow-up visit, and PFS was calculated from the date of first chemotherapy to the date of progression. OS and PFS were estimated using the Kaplan-Meier product-limit method. Survival rates were compared for statistical differences using log-rank analysis. A prognostic model was established by identifying all variables that significantly influenced OS or PFS at a p-value less than or equal to 0.05 in univariate analysis. Multivariate analysis was performed using stepwise Cox proportional hazards regression modeling. Factors included in the multivariate analyses were the following: age, sex, performance status, previous esophagectomy, weight loss (10% of weight lost for 3 months) at the time of chemotherapy, stent or percutaneous endoscopic gastrostomy, tracheoesophageal fistula, primary tumor location, histologic tumor grade, and metastatic site (lung, liver, bone, and adrenal gland). In addition, laboratory findings at the time of chemotherapy were included as categorical variables based on the normal ranges and consisted of the following: hemoglobin (>11.2 g/dL vs. ≤11.2 g/dL), white blood cell count (>8,630/µL vs. ≤8,630/µL), platelet count (> 138×103/µL vs. ≤138×103/µL), serum albumin (>3.5 g/dL vs. ≤3.5 g/dL), total bilirubin (>1.5 mg/dL vs. ≤1.5 mg/dL), aspartate aminotransferase (>40 U/L vs. ≤40 U/L), alanine aminotransferase (>40 U/L vs. ≤40 U/L), alkaline phosphatase (>128 U/L vs. ≤128 U/L), and serum creatinine (>1.1 mg/dL vs. ≤1.1 mg/dL).
A nomogram was developed as a prognostic scoring system incorporating significant clinical and laboratory variables based on a multivariate Cox proportional hazards regression model. Points for variables in the nomogram were determined by their Cox regression model coefficients. Variables were selected based on their statistical significance (p<0.05) in multivariate analysis. For example, the predicted 12-month OS hazard rate was calculated as follows: 12 months OS hazard rate=0.06501×exp (0.6755×ECOG 0-1 [if the ECOG grade was 0-1, the score could be 1, if not, the score is 0])+1.0682×ECOG 2 (if the ECOG grade1 was 2, the score could be 1, if not, the score is 0)+0.3784×weight loss (absence of weight loss=0, presence of weight loss=1)-0.6562×albumin (if serum albumin >3.5 g/dL, the score is 1, if serum albumin ≤3.5 g/dL, the score is 0)-0.2848×previous esophagectomy (absence=0, presence=1). The nomogram was then constructed using coefficients from the Cox proportional hazard model for convenience.
On the basis of the nomogram, patients were categorized into three groups according to their risk score percentile: favorable (0-50 percentile), intermediate (50-75 percentile), and poor (75-100 percentile) risk score. Performance of the nomogram was evaluated using discrimination and calibration. Discrimination was examined using the area under the receiver operating characteristic (ROC) curve (AUC). For internal validation, random sampling was repeated with replacement using the bootstrapping method over 1,000 times, and a new bias-corrected 95% confidence interval (CI) for AUC was calculated using R 2.13.2 (R Foundation for Statistical Computing, Vienna, Austria; ISBN 3-900051-07-0;
Between January 2000 to December 2010, 239 MESCC patients were treated with either FP or XP chemotherapy. Patient characteristics are provided in
The baseline characteristics of the validation patient set are also provided in
Univariate analyses revealed that poor ECOG score, weight loss (loss of at least 10% of weight for 3 months), absence of previous esophagectomy, low hemoglobin level (≤11.2 g/dL), and low albumin level (≤3.5 g/dL) were significantly associated with low OS. Moreover, poor ECOG score, weight loss, absence of previous esophagectomy, low albumin level, leukocytosis (>8,630/µL), elevated alanine aminotransferase level, and the presence of liver metastasis predicted short PFS in univariate analysis (
To build a nomogram to predict survival following frontline palliative chemotherapy in MESCC, a scoring system was constructed that incorporates significant variables such as performance status scores, weight loss, low albumin levels, and previous esophagectomy (
All 239 patients were scored according to the nomogram. Patients were then categorized into three groups by risk scores based on the nomogram: favorable (score 0-90, n=119), intermediate risk (score 91-135, n=58), and poor (score > 135, n=61) risk score (
The ROC curve for 12-month OS is shown in
Very little research has been done on the role of palliative chemotherapy in MESCC. This is largely because of the absence of largescale phase III randomized clinical trials, since MESCC is included as part of esophageal adenocarcinoma/gastric cancer in most trials. Despite the lack of clear evidence that palliative chemotherapy confers a survival benefit in MESCC, many physicians are firmly convinced that chemotherapy has a favorable impact on outcome because some drugs (or combination of cytotoxic agents) have shown antitumor activity in phase II trials [
Only a few studies to date have focused on prognostication of MESCC with palliative chemotherapy [
In our study, we found that poor performance status (ECOG≥2), low albumin level (≤3.5g/dL), absence of previous esophagectomy, and weight loss (at least 10% for 3 months) significantly predicted poor survival and short PFS following first-line fluoropyrimidine/cisplatin chemotherapy. It is worth noting that all of these factors, although independent prognosticators, are more or less related to nutritional matters in esophageal cancer. Hence, the actual benefit from chemotherapy in MESCC patients with poor nutritional status as indicated by low albumin, poor performance or weight loss needs to be taken into consideration during decision making. The greatest advantage of our nomogram is that it is based on clinical parameters that are prospectively collected before chemotherapy. The robustness of the model was demonstrated using internal validation via the bootstrapping method as well as external validation on a series of French patients with MESCC who received an FP regimen. Our study is also the largest study analyzing the prognostic factors of outcome in MESCC treated with a fluoropyrimidine/cisplatin regimen, the most commonly used and most efficient regimen according to a review of 96 different trials performed by Grunberger et al. [
The nomogram including the variables, ECOG performance, albumin, weight loss and previous esophagectomy predicts survival in MESCC patients and could serve as a guide for the use of FP/XP chemotherapy in MESCC patients.
Conflict of interest relevant to this article was not reported.
Nomogram for overall survival. ECOG, Eastern Cooperative Oncology Group; OS, overall survival.
Survival curve according to nomogram score. OS, overall survival; PFS, progression-free survival.
(A) Receiver operating characteristic curve for 12-month overall survival. (B) Calibration curve for predicted 12-month survival probability. Axis X, predicted 12-month survival probability; axis Y, observed 12-month survival probability; dashed line, ideal reference line; solid line, current nomogram performance with 95% confidence intervals; circle, median observed survival probability; x, median observed survival probability after bias correction. ROC, receiver operating characteristic; OS, overall survival; XP, capecitabine/cisplatin; AUC, receiver operating characteristic curve; FP, fluorouracil/cisplatin.
Survival curves (overall survival [OS] and progression-free survival [PFS]) according to the nomogram score in the validation set.
Patient characteristics
Characteristic | Training set |
Validation set |
p-value |
---|---|---|---|
Age | 0.549 | ||
Median age (range) | 62 (40-79) | 56 (42-77) | |
≤70 | 204 (85.4) | 54 (88.5) | |
>70 | 35 (14.6) | 7 (11.5) | |
Gender | 0.478 | ||
Male | 230 (96.2) | 57 (93.4) | |
Female | 9 (3.8) | 4 (6.6) | |
ECOG | ≤0.001 | ||
0-1 | 225 (94.1) | 42 (68.9) | |
>2 | 14 (5.9) | 19 (31.1) | |
Previous esophagectomy | ≤0.001 | ||
Yes | 93 (38.9) | 9 (14.8) | |
No | 146 (61.1) | 52 (85.2) | |
Histology grade | 0.063 | ||
WD | 12 (5.0) | 13 (28.3) | |
MD | 160 (66.9) | 15 (32.8) | |
PD | 35 (14.6) | 18 (39.1) | |
Unknown | 32 (13.5) | 15 (24.6) | |
Lung metastasis | 0.003 | ||
Presence | 97 (40.6) | 38 (62.3) | |
Absence | 142 (59.4) | 23 (37.7) | |
Liver metastasis | 0.002 | ||
Presence | 36 (15.1) | 20 (32.8) | |
Absence | 203 (84.9) | 41 (67.2) | |
Bone metastasis | 0.188 | ||
Presence | 26 (10.9) | 11 (18.0) | |
Absence | 213 (89.1) | 50 (82.0) | |
Adrenal gland metastasis | 0.743 | ||
Presence | 13 (5.4) | 2 (3.4) | |
Absence | 226 (94.6) | 57 (96.6) |
Values are presented as number (%). ECOG, Eastern Cooperative Oncology Group; WD, well differentiated; MD, moderately differentiated; PD, poorly differentiated.
Univariate analysis for overall survival and progression-free survival
Variable | Median OS (mo) | p-value | Median PFS (mo) | p-value |
---|---|---|---|---|
Age (yr) | 0.12 | 0.47 | ||
≤70 | 12.2 | 4.7 | ||
>70 | 9.9 | 4.1 | ||
Gender | 0.99 | 0.41 | ||
Male | 11.2 | 4.6 | ||
Female | 10.5 | 5.4 | ||
ECOG | <0.001 | 0.03 | ||
0-1 | 11.3 | 4.7 | ||
>2 | 4.5 | 2.1 | ||
Weight loss (10% in 3 months) | <0.001 | 0.004 | ||
Yes | 9.1 | 3.5 | ||
No | 12.7 | 4.9 | ||
Obstructive symptom | 0.50 | 0.88 | ||
Yes | 11.1 | 4.9 | ||
No | 11.2 | 4.4 | ||
Stent or PEG | 0.04 | 0.13 | ||
Yes | 10.5 | 3.2 | ||
No | 11.2 | 4.8 | ||
TEF | 0.11 | 0.36 | ||
Yes | 7.0 | 5.0 | ||
No | 11.2 | 4.6 | ||
Upper E | 0.71 | 0.93 | ||
Yes | 9.6 | 4.4 | ||
No | 11.3 | 4.7 | ||
Previous esophagectomy | <0.001 | 0.002 | ||
Yes | 13.8 | 5.7 | ||
No | 10.0 | 4.4 | ||
Histology grade | 0.26 | 0.93 | ||
WD | 8.9 | 4.6 | ||
MD | 11.3 | 4.8 | ||
PD | 9.1 | 4.1 | ||
WBC (/μL) | 0.08 | 0.06 | ||
≤8,630 | 11.3 | 5.0 | ||
>8,630 | 10.3 | 3.1 | ||
Hb (pre-transfusion, g/dL) | 0.02 | 0.20 | ||
≤11.2 | 9.1 | 4.3 | ||
>11.2 | 12.8 | 4.9 | ||
Platelet (/μL) | 0.89 | 0.37 | ||
≤138×103 | 11.1 | 6.9 | ||
>138×103 | 11.1 | 4.6 | ||
Albumin (g/dL) | <0.001 | <0.001 | ||
≤3.5 | 7.1 | 3.2 | ||
>3.5 | 13.0 | 5.4 | ||
Total bilirubin (mg/dL) | 0.43 | 0.95 | ||
≤1.5 | 11.2 | 4.6 | ||
>1.5 | 17.6 | 5.9 | ||
AST (U/L) | 0.18 | 0.17 | ||
≤40 | 11.3 | 4.8 | ||
>40 | 5.9 | 3.8 | ||
ALT (U/L) | 0.16 | 0.07 | ||
≤40 | 11.7 | 4.8 | ||
>40 | 8.3 | 3.3 | ||
ALP (U/L) | 0.53 | 0.17 | ||
≤128 | 11.3 | 4.7 | ||
>128 | 10.2 | 4.3 | ||
Creatinine (mg/dL) | 0.33 | 0.20 | ||
≤1.1 | 11.3 | 4.6 | ||
>1.1 | 10.5 | 4.4 | ||
Lung metastasis | 0.15 | 0.98 | ||
Presence | 13.5 | 4.6 | ||
Absence | 10.8 | 4.7 | ||
Liver metastasis | 0.11 | 0.01 | ||
Presence | 10.9 | 4.1 | ||
Absence | 11.1 | 4.7 | ||
Bone metastasis | 0.27 | 0.37 | ||
Presence | 9.6 | 3.8 | ||
Absence | 11.3 | 4.8 | ||
Adrenal gland metastasis | 0.74 | 0.54 | ||
Presence | 12.7 | 4.4 | ||
Absence | 11.1 | 4.7 |
ECOG, Eastern Cooperative Oncology Group; PEG, percutaneous endoscopic gastrostomy; TEF, tracheoesophageal fistula; Upper E, involvement of upper esophagus; WD, well differentiated; MD, moderately differentiated; PD, poor differentiated; WBC, white blood cell; Hb, hemoglobin; AST, aspartate aminotransferase; ALT, alanine aminotransferase; ALP, alkaline phosphatase.
Multivariate analysis for overall survival and progression-free survival
Characteristic | Overall survival |
Progression free survival |
||||
---|---|---|---|---|---|---|
p-value | RR (exp. B) | 95% CI | p-value | RR (exp. B) | 95% CI | |
Poor performance ECOG>2 | 0.01 | 2.27 | 1.1-4.3 | 0.30 | 1.4 | 0.7-2.5 |
Weight loss (10% in 3 mo) | 0.04 | 1.43 | 1.0-2.0 | 0.10 | 1.3 | 0.9-1.7 |
Previous esophagectomy | <0.01 | 0.63 | 0.5-0.9 | 0.03 | 0.7 | 0.5-0.9 |
Albumin less than 3.5 g/dL | <0.01 | 0.58 | 0.4-0.8 | <0.01 | 0.7 | 0.5-0.9 |
Hb less than 11.2 g/dL | 0.06 | 0.7 | 0.5-1.0 | 0.17 | 1.3 | 0.9-1.9 |
History of stent or PEG | 0.96 | 0.98 | 0.6-1.5 | |||
Presence of liver metastasis | 0.17 | 1.3 | 0.9-1.9 |
RR, relative risk; CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; Hb, hemoglobin; PEG, percutaneous endoscopic gastrostomy.
Prognostic risk groups based on nomogram score
Nomogram score | Median OS (mo) | p-value | 95% CI | Median PFS (mo) | p-value | 95% CI |
---|---|---|---|---|---|---|
0-90 | 13.8 | <0.001 | 10.8-18.6 | 5.9 | <0.001 | 4.6-7.2 |
91-135 | 11.2 | 8.7-11.9 | 4.3 | 3.3-5.3 | ||
>135 | 7.0 | 3.6-10.0 | 3.0 | 2.3-3.8 |
OS, overall survival; CI, confidence interval; PFS, progression-free survival.