Predictive factors of survival in patients treated with definitive chemoradiotherapy for squamous cell esophageal carcinoma

Abstract

AIM: The aim of the study was to evaluate the predictive factors of survival in patients with locally advanced squamous cell esophageal carcinoma (LASCOC) treated with definitive chemoradiotherapy (CRT) regimen based on the 5FU/CDDP combination.

METHODS: All patients with LASCOC treated with a definitive CRT using the 5FU/CDDP combination between 1994 and 2000 were retrospectively included. Clinical complete response (CCR) to CRT was assessed by esophageal endoscopy and CT-scan 2 mo after CRT completion. Prognostic factors of survival were assessed using univariate and multivariate analysis by the Cox regression model.

RESULTS: A total of 116 patients were included in the study. A CCR to CRT was observed in 86/116 (74.1%). The median survival was 20 mo (range 2-114) and the 5-year survival was 9.4%. Median survival of responder patients to CRT was 25 mo (range 3-114) as compared to 9 mo (range 2-81) in non-responder patients (P < 0.001). In univariate analysis, survival was associated with CCR (P < 0.001), WHO performance status < 2 (P = 0.01), tumour length < 6 cm (P = 0.045) and weight loss < 10% was in limit of significance (P = 0.053). In multivariate analysis, survival was dependant to CCR (P < 0.0001), weight loss < 10% (P = 0.034) and WHO performance < 2 (P = 0.046).

CONCLUSION: Our results suggest that survival in patients with LASCOC treated with definitive CRT was correlated to CCR, weight loss and WHO performance status.

Key Words: Definitive chemoradiotherapy, Esophageal squamous cell carcinoma, Predictive factors

INTRODUCTION

Esophageal cancer is a frequent gastrointestinal malignancy with 32 332 new cases per year in Europe. In France, esophageal carcinoma is the third most frequent digestive tract cancer with approximately 5000 new cases per year[1,2]. Approximately 50% of patients present a locally advanced esophageal carcinoma at diagnosis. To date, the incidence rate of adenocarcinoma is increasing but squamous cell carcinoma still remains the most frequent histological type in France[1,2]. The definitive chemoradiotherapy (CRT) based on the Herskovic regimen is considered as the standard medical treatment in non operated patients with locally advanced esophageal carcinoma[3,4]. Moreover, two phase III trials recently suggested that definitive CRT could be considered as an alternative treatment in patients with esophageal carcinoma[5,6].

However, some questions remain unsolved as regards the CRT regimen optimisation. Moreover, although most studies included both patients with squamous cell and adenocarcinoma, it has been suggested to consider these two tumours separately for treatment as regards their different risk factors, carcinogenesis pathways and treatment response[3-5,7-11]. Furthermore, the study of Rizk et al recently reported that long term prognosis in patients with esophageal carcinoma treated with preoperative CRT could be significantly different according to the histological type of tumour[11]. However, few reported series have specifically focused on the long-term survival analysis in patients with squamous cell carcinoma treated with definitive CRT using the 5FU/CDDP combination[12-20]. Moreover, most of these series included a limited number of patients or reported results with short follow-up[13-20].

The aim of the present study was to assess the long-term results and the predictive factors of survival in a large series of patients with a locally advanced squamous cell esophageal carcinoma (LASCOC) treated with a definitive CRT regimen based on the 5FU/CDDP chemotherapy (CT) combination. Furthermore, knowledge of these prognostic factors could be useful for the management of individual patients as well as a stratification variable for the design of future randomised trials.

MATERIALS AND METHODS

Patient population

All consecutive patients with a LASCOC referred between January 1994 and the 31^st December 2000 were retrospectively included for the study. Patients were selected based on the following criteria: a histologically confirmed squamous cell carcinoma; a first-line treatment with a definitive CRT regimen using the 5FU/CDDP CT and concomitant external radiotherapy (RT). Patients were excluded if they had previous a history of carcinoma during the past three years and if they had synchronous distant metastasis. In our centres, the definitive CRT regimen based on the 5FU/CDDP combination was the first therapeutic option used in patients with LASCOC.

For each patient, we routinely recorded all baseline clinical and tumour characteristics including age, sex, World Heath Organisation (WHO) performance status, dysphagia Atkinson score and weight loss at the beginning of treatment, median tumour length, esophageal tumour location, and tumour stage. Events and toxicity related to treatment were also included in the computer data base.

Tumour stage

The 1983 AJCC staging system was used in this study according to recently published recommendations[21]. Tumour evaluation was based on oesophagoscopy, barium oesophagography, chest and abdominal computed tomography (CT-scan), endoscopic bronchoscopy and esophageal ultrasonography when feasible.

Treatment schedule

CRT regimen was based on the 5 FU/CDDP CT combination associated with an external RT. The RT was delivered either by a dose of 50 Gy (50 Gy/25 fractions per 5 wk) with concomitant CT courses delivered on wk 1 and 5, or either a dose of 60 Gy (20 Gy/10 fractions × 3 courses separated by a 2-wk break) with concomitant CT courses delivered on wk 1, 5 and 9. The CT courses combined 5-FU (750 to 1000 mg/m² per day delivered by continuous infusion on 4 d) and CDDP (75 to 100 mg/m² delivered on 1 d). The target volume of RT was the macroscopic tumour and enlarged lymph nodes, if any, surrounded by 5 cm proximal and distal margins and a 2 cm radial margin. The target was extended to the inferior cervical area in cases of tumours located above the carina. The specified dose was delivered at the intersection of the central axis of the beams, according to international guidelines. The irradiation technique was applied in anterior and posterior opposed fields. At 40 Gy, the radiation portals were reduced to shield the spinal cord and encompass the primary tumour with a 2-3 cm craniocaudal margin.

Evaluation of clinical response and toxicity to CRT

Patients were considered to have a clinical complete response (CCR) to CRT when no residual tumour was identified on endoscopy and when no metastatic disease occurrence was observed on CT-scan. This evaluation was performed 2 mo after CRT completion.

Toxicity related to the treatment was evaluated using the National Cancer Institute Common Toxicity Criteria (NCI-CTC, version 2.0). Toxicity was assessed in each patient at d 1 of each chemotherapy course. At each course, patients received the treatment when they exhibited a WHO performance status of 2 or less; satisfactory haematologic function (leucocytes count ≥ 3000 mm^-3, platelet count ≥ 100 000 mm^-3) and good renal function (creatine serum level ≤ 100 micromole/L). Patients with major complication i.e. heart disease, pulmonary fibrosis, or active carcinoma at the other site were not eligible for treatment.

Follow-up

The follow-up was performed on clinical basis, endoscopy and CT scan. Histopathological confirmation of the recurrence was not routinely required. Follow-up was performed either until death or for the purpose of this study until October 2005.

Statistical analysis

Analysis was performed in October 2005 and was considered the cut-off date. Survival curves according to the putative prognostic factors were established using the Kaplan-Meier method and were compared with a log-rank test. The effects of clinical characteristics at baseline related to prognosis using univariate analysis were further evaluated in multivariate analysis using Cox regression model. A two-side P-value equal or less than 0.05 was considered to indicate statistical significance. Data from patients who had been lost to follow-up were censored at the time of last obtained information. The date of CRT initiation was the starting point for the analysis of overall survival. The date of CRT response evaluation was the starting point for the analysis of the disease free survival.

RESULTS

Patients characteristics

Between the first of January 1994 and the 31^st December 2000, one hundred and sixteen consecutive patients were treated with a CRT based on the 5FU/CDDP CT combination. The majority of patients had a good performance status and the dysphagia score prior to CRT reflected their ability to eat a normal or semisolids diet for approximately 90% of these patients (Table 1). Among patients who were estimated with T1-T2 tumour on CT-scan, 14 were estimated to present with a T1-T2 N0 tumour. These latter patients were treated with a definitive CRT as regards age and/or comorbidities.

Table 1 Patient characteristics.

	n = 116	%
Mean age (yr)	61.3 (40-90)
Male	101	87.1
Female	15	12.9
WHO performance status (OMS)
0	32	27.6
1	70	60.3
2	14	12.1
Dysphagia (Atkinson score)
0	2	1.7
1	21	18.1
2	58	50
3	26	22.4
4	9	7.8
Weight loss ≥ 10% at CRT
Initiation	30	25.9
TNM
T 1-2	27	23.3
3-4	89	76.7
N 0	50	43.1
1	54	46.5
x	12	10.4
M 0	116	100
1	-	-
Esophageal location
Upper one-third	35	30.2
Middle one-third	53	45.7
Lower one third	28	24.1
Mean tumour length (cm)	4.9 (0-15)
Histopathology
Squamous cell carcinoma	116

Safety and toxicity per patient

Significant toxicities per patient are shown in Table 2. There was no death related to CRT. Moreover, 75 patients (64.6%) experienced grade 3-4 toxicities and 112 (96.5%) patients achieved the planned concomitant CRT regimen. Dose modification to the planned CT regimen were required in 55 patients (47.4%) and 18 patients (15.5%) received at least one of their CT courses with a delay of more than 1 wk. The mean delivered radiation dose was 53.2 Gy.

Table 2 Significant treatment toxicities per patient (%).

	Grade 3 (%)	Grade 4 (%)
Haematological
Neutropenia	9	3
Anemia	11	-
Thrombopenia	1	1
Mucositis
Oral	4	-
Esophageal	16	-
Gastrointestinal
Nausea	12	-
Diarrhoea	4	-
Neuropathy	1	-
Alopecia	3	-

During the CRT treatment, 18 patients (15.5%) required nutritional enteral feeding. In contrast, palliation of the dysphagia by endoscopic stenosis dilation was performed in 25 patients (21.6%). A self expandable metallic stent was inserted in 6 patients (5.1%) during CRT treatment.

Clinical complete response (CCR) to CRT

A total of 86/116 patients (74.1%) achieved a CCR to CRT. In the remaining 30 non-responder patients, a self expanding metallic stent was inserted in 12 for dysphagia palliation after completion of CRT, a CT treatment was initiated in 1 patient and salvage surgery was performed in 2 patients.

Patient outcome

In October 2005, 7/116 patients (6%) were still alive. The median follow-up of these surviving patients was 79 mo (range 56-104) and the median follow-up of the 11 patients who were lost to follow-up during the study was 45.1 mo (range 9-68).

The median overall survival was 20 mo (range 2-114) and the 2-years and 5-years survival rates were 39.6% and 9.4%, respectively. Moreover, the median overall survival of the 86 patients who had CCR to CRT was 25 mo (range 3-114) as compared to 9 mo (range 2-81) in non-responder patients (P < 0.001) (Figure 1). The median disease free survival of responder patients to CRT was 17 mo.

Figure 1 Survival according to response to CRT. The median overall survival of patients who had a complete clinical response (CCR) to the chemoradiotherapy (CRT) was 25 mo as compared to 9 mo in non-responder patients (P < 0.001).

During the follow-up, 34 of responder patients (39.5%) experienced a local disease recurrence, 37 patients (43%) experienced metastatic disease and 19 of them experienced both of these recurrences.

Prognostic factors of survival

As regards univariate analysis (Table 3), survival was correlated to CCR to CRT (P < 0.001), WHO performance status < 2 (P = 0.01) and tumour length < 6 cm (P = 0.045). In contrast, weight loss > 10% at the start of CRT was in limit of statistical significance and was included in multivariate analysis (P = 0.053). In a Cox regression model (Table 3), the independent covariates significantly associated with survival were the CCR to CRT (P < 0.0001; Odds Ratio (OR): 0.121; IC95 = 0.06-0.24), the weight loss < 10% (P = 0.034; OR: 0.53; IC95 = 0.29-0.95) and a WHO performance status < 2 (P = 0.046; OR: 0.495; IC95 = 0.24-0.99).

Table 3 Predictive factors of survival, univariate and multivariate analysis.

	Univariate	Multivariate
	P	P	OR	IC95
Sex	0.507	-
Age < 70	0.745	-
WHO performance status < 2	0.01	0.046	0.52	0.28-0.99
Weight loss < 10%	0.053	0.034	0.53	0.29-0.95
Dysphagia	0.074	-
T	0.273	-
N	0.499	-
Tumour location	0.501	-
Tumour length < 6 cm	0.045	0.534	0.86	0.53-1.38
Complete response to CRT	< 0.001	< 0.000	0.21	0.13-0.36

DISCUSSION

To date, definitive CRT based on the 5FU/CDDP combination is considered as standard treatment in non operable patients with locally advanced esophageal carcinoma whatever the histological tumour phenotype[3-5]. Some authors recently suggested that histological types of oesophagus tumour could be considered separately regarding their significant different treatment response and long term prognosis[9-11]. Therefore, we performed a retrospective analysis of the long term outcome and predictive factors of survival in 116 patients with LASCOC treated with a definitive CRT using the 5FU/CDDP CT combination.

In our study, the 5-year survival was 9.4% and the median overall survival was 20 mo. We also found that responder patients to CRT had a significantly increased median survival as compared to non-responders patients (24 mo vs 9 mo; P < 0.001). This result was supported by the multivariate analysis which identified the CCR as an independent prognostic factor of survival. In definitive CRT series using the 5FU/CDDP combination, a survey of literature showed that median overall survival ranged from 17 to 26 mo and the 5-year survival rate from 20% to 30%[12-20]. The 5-year survival rate in our study was slightly lower as compared to those reported in these series. This result could be explained by the patient selection bias in these prospective trials, whereas our retrospective study possibly reflected the outcome of non-selected patients with LASCOC treated with definitive RT.

The CCR to CRT was obtained in 75.9% of patients in our series. Moreover, a CCR was identified as an independent prognostic factor of long-term survival in our multivariate analysis. Although the prognostic significance of pathological complete response after preoperative CRT was well documented[22-24], to our knowledge, there are no previous studies that have reported similar result in patients with LASCOC treated with definitive CRT using the 5FU/CDDP combination. In fact, the significant impact of CCR to CRT in long-term survival in patients treated with the same definitive CRT regimen was reported in series which included patients with mixed histological tumour types[12,25,26]. Moreover, in the reported Ohtsu et al study focusing exclusively on patients with LASCOC, the CCR to CRT was identified as a predictive factor of the progression free survival but not for overall survival[18]. In our study, 39.5% of responder patients to CRT had a local disease recurrence. In previous studies, local recurrences were reported to be as high as 38% to 48% after definitive CRT[12,18]. Furthermore, a distant metastasis occurred in 43% of responder patients to CRT in our series. This result compared less favourably to other series including patients with squamous cell carcinoma treated with the same CT combination, where less than 30% of responder patients to CRT experienced a distant metastasis[12,18]. The frequent use of additional CT in responder patients to CRT in these latter series could probably explain the difference in metastasis frequency. Indeed, only 44% of patients received additional CT after the CRT completion in our study.

Thus, our results suggest that further optimisations of the definitive CRT regimen are required for both local and systemic disease control improvement. To date, the 5FU/CDDP combination used in our study is still considered the standard for the CRT regimen. Evaluation of novel chemotherapy regimens which include new drugs such as irinotecan and new cancer therapies encompassing those directed against vascular growth factor and epidermal growth factor pathways may be usefully associated with RT to provide improved sustained CCR and therefore optimal long-term survival[27-30]. Furthermore, CRT regimen optimisation should determine the optimal radiotherapy dose in order to achieve a sustained local disease control. In our study, patients received a mean dose of 52.1 Gy of external RT in the tumour bed, which was a similar dose as that used in the standard regimen described by Herskovic et al[3]. Recently, Zhang et al reported, in a retrospective study, that patient who received a dose of RT more than 51 Gy had a statistically better local disease-free survival and overall survival as compared to patients treated with a dose of RT less than 51 Gy[31]. Minsky et al specifically investigated dose escalation in RT. In this prospective study, patients were randomised in a CRT regimen using either 50.4 Gy of RT or 64.8 Gy of RT[32]. However, the unexplained excess of death frequency (9%) which occurred prior to the dose escalation in patients treated in the 64.8 Gy arm did not permit a valid conclusion as regards the optimal dose of RT. In our study, no death due to CRT was observed. However, we found that 64.6% of patient experienced grade 3-4 toxicities, which was a similar result to that usually reported[3,4]. As regards the new RT techniques including three dimensional CT planning with conformal beam, dose escalation in the definitive CRT regimen could be reconsidered. Further studies including these new RT techniques may be helpful to evaluate the optimal and safety dose of RT in patients with LASCOC.

Multivariate analysis identified the WHO performance status less than 2 as the second independent predictive factor of long-term survival in patients treated with definitive CRT for a LASCOC. The WHO performance status has been previously identified as prognostic factor in patients with esophageal cancer treated by radiotherapy alone or in patients with metastatic oesophago-gastric cancer in previous studies[21,33]. Moreover, Polee et al also identified this variable as a prognostic factor of survival in a meta-analysis performed in six prospective trials which included several CRT regimens and mixed histological type of esophageal tumour[34]. However, to our knowledge, this result had never been previously reported in patients with LASCOC treated in curative intent with definitive CRT using the 5FU/CDDP combination. The WHO performance status evaluation provides useful information on the patient’s general well-being indirectly reflecting the impact of the malignancy on the physiological individual condition. Although the evaluation of the WHO performance status seems to be subjective, our results showed that this variable was significantly correlated with survival. Moreover, weight loss before treatment starting was also identified as prognostic factor in our work. This variable was recently to be correlated with prognosis in a meta-analysis by Thomas et al in patient treated by definitive CRT for esophageal cancer[30].

Interestingly, tumour characteristics were not identified as prognostic factors in our study. Indeed, the tumour diameter and the tumour length were only significant in univariate analysis. In surgical series, the extent of the tumour infiltration in the esophageal wall and the lymph node involvement were frequently closely correlated with survival[23,24,35]. Although an underpowered significance could not be excluded in the analysis, our results could also reflect the limited accuracy of the usual staging evaluation including oesophagography, oesophagoscopy and thoraco-abdominal computed tomography used during the period of the study. However, these imaging modalities appear to be more accurate for the CRT response assessment as regards the significant link that was observed between the CCR and survival. The high rate of early disease recurrence in our study could reflect an overestimation of the CCR rate by modalities used in our study. Thus, the further use of endoscopic ultrasound and PET-FDG, may be helpful to provide better correlative data for initial tumour staging and for tumour response assessment[36-38].

In conclusion, based on a large series of patients with LASCOC, our results suggest that survival of these patients treated with definitive CRT using the 5FU/ performance status. These prognostic factors could be considered for the management of individual patients as well as a stratification variable for the design and interpretation of further randomised trials. However, as regards the retrospective design of our study, further prospective studies are necessary to investigate the impact of these prognostic factors.

ACKNOWLEDGMENTS

The authors thank Richard Medeiros, Rouen University Hospital medical Editor, for his valuable advice in editing the manuscript.