This article is mentioned in:

Introduction

Esophageal cancer is the eighth most common type of cancer worldwide, with >480,000 new cases diagnosed annually (1). Esophageal cancer has a high mortality rate (sixth worldwide), causing >400,000 deaths annually (2). Squamous cell carcinoma is the most frequently type occurring in Asians, particularly in China, where it accounts for 70% of global morbidity (3). However, the incidence of esophageal adenocarcinoma in Western populations is rapidly increasing, whereas that of squamous cell carcinoma remains unchanged (4). Esophagectomy is considered to be the standard treatment for patients with resectable esophageal carcinoma, despite a detailed assessment of preoperative staging showing that 25% of patients treated with definitive surgery had microscopically positive resection margins (R1). However, the 5-year survival rate scarcely exceeds 40% (5); in addition, due to the morbidity and mortality associated with surgery, this approach is limited to a minority of medically fit patients.

Since the 1980s, there have been several randomized clinical trials (RCTs) assessing the efficacy of preoperative chemoradiotherapy followed by surgery (CRTS) in the treatment of esophageal cancer. However, the sample-size of these RCTs was small, with a short-term follow-up and adverse outcomes in the surgical monotherapy arm of combination treatment trials when compared with surgery alone (SA) case-series (6). Furthermore, the majority of the trials did not have sufficient statistical power to produce a definitive conclusion. Thus, a comprehensive analysis was conducted to compare the potential objective value of CRTS with SA for resectable esophageal carcinoma.

As regards the differences between traditional and cumulative meta-analysis, cumulative meta-analysis refers to a meta-analysis of the obtained studies in a certain order; those studies are treated as a continuous whole and multiple meta-analyzes are performed by accumulating studies sequentially in a specified sequence (such as publication time). In addition, if a new test result is published, a new meta-analysis may follow. Traditional meta-analysis is performed only once, whereas cumulative meta-analysis is performed several times; the former may obtain summary results, but cannot distinguish the impact of each study result on the summary results, whereas the latter does not only obtain the results of the summary and compare the dynamic results of summary changes, but also compares the effect of the newly added studies on overall outcome. The cumulative meta-analysis is controversial in terms of test level. Some scholars object to performing multiple meta-analyses due to the increasing probability of committing class I error, and claim the test level should be adjusted for each analysis; some scholars believe that the analysis of the Bayesian theory may be used to explain, without the need for adjustment.

Based on this theory, the present study aimed to combine the traditional and cumulative meta-analysis to explore the pooled results of the relevant studies.

Data collection methods

Search strategy

The relevant articles identified were RCTs retrieved from Embase, PubMed and The Cochrane Library (issue 4, 2016) and the deadline for trial publication and/or presentation was October 1st, 2016. The American Society of Clinical Oncology (ASCO) and the Cochrane Collaboration's Central Register of Controlled Clinical Trials were searched for updates of the trials. The search terms were as follows: Esophageal neoplasms, esophageal cancer, esophageal carcinoma, esophageal tumor, neoadjuvant therapy, chemoradiotherapy, esophagectomy, resection, surgery and operation.

Inclusion and exclusion criteria

In our meta-analysis, the study focus was locoregional resectable esophageal cancer patients who received either CRTS or SA. The eligible studies were required to meet the following inclusion criteria: i) Prospective RCTs comparing CRTS vs. SA in the initial management of resectable esophageal cancer; ii) outcome indices containing survival data; iii) no significant differences in baseline characteristics between the CRTS and SA groups; and iv) definitive follow-up survival number of cases or survival curve, with a follow-up rate of >95% in the original RCTs. Studies focusing on patients with esophageal cancer who had been treated with neoadjuvant chemotherapy alone or radiotherapy alone, other studies without usable data, letters, editorials, case reports and reviews were excluded.

Data extraction and quality assessment

Two investigators independently extracted data to avoid bias in the course of the extraction. Disagreements were resolved by consensus or consultation with third parties. Statistics for each available outcome were extracted from trials in the light of the key information including patient characteristics, first author, year of publication, country/region, the regimen of the CRTS, and tumor histology. The methodological quality assessment of individual studies followed the Cochrane risk of bias method.

Statistical analysis

Overall survival rates at 1, 3 and 5 years (OSR1y, OSR3y and OSR5y, respectively), R0 resection rate, postoperative mortality, postoperative local recurrence rate and postoperative distant metastasis rate were extracted and pooled with 95% confidence intervals (CIs) by adopting the fixed- or random-effects model where heterogeneity was assessed with the inconsistency statistic (I2<50%, P>0.05; and I2≥50%, P≤0.05, respectively). The odds ratio (OR) was estimated with 95% CI and P-values in both the CRTS and SA groups. All calculations were performed using Review Manager 5.3 (Nordic Cochrane Centre, Copenhagen, Denmark), R software version 3.2.2, and STATA version 12.0 (StataCorp LP-College Station, TX, USA). Statistical significance was set at P<0.05.

Results

Summary of included studies

A total of 1279 records were identified according to the search strategy and 22 were finally included in the meta-analysis after removing duplicated, ineligible and unrelated studies (Fig. 1). Ten countries, including China, Australia, Japan, Korea, Thailand, United States of America, France, The Netherlands, Ireland and Norway, were included in the RCTs. Of the 22 studies, 20 (7–26) reported OSR1y, 19 (7–21,24–27) reported OSR3y and 15 (8,11–13,15–17,19–12,24–27) reported information on OSR5y after SA or CRTS for resectable esophageal carcinoma.

Figure 1. Trial selection process.

As regards pathological type, 5 RCTs (10,15,24,26,27) on adenocarcinoma and 12 studies (7–9,11,13,14,16,21,23,24,26) on squamous cell carcinoma investigated OSR1y, OSR3y and OSR5y after CRTS or SA for resectable esophageal carcinoma.

A total of 9 studies (7,9,11,13–15,23,27,28) reported R0 resection rate, 10 (9,12–15,16,19–21,26) included postoperative local recurrence rate and distant metastasis rate, and 15 (7–12,14–16,20,21,28) provided postoperative mortality information.

With respect to the treatment efficacy of both methods in different countries or regions, 8 trials (8,13,14,16–19,23) collected data from Asian populations, 7 studies (7–11,24–26) from European populations, and 2 (12,20) from USA populations. The study characteristics are summarized in Table I.

Table I. Outline of the literature search findings and the general characteristics of the included studies.

Table I.

Outline of the literature search findings and the general characteristics of the included studies.

Study, year	Country	Chemotherapy	Radiotherapy	Pathology	CRTS/SA, n	(Refs.)
Nygaard et al, 1992	Norway	2 cycles: Cisplatin 20 mg/m2 D1-5; bleomycin 5 mg/m2 D1-5	35 Gy, 1.75 Gy/fr, 4 weeks	SCC	47/41	(7)
Apinop et al, 1994	Thailand	2 cycles: Cisplatin 100 mg/m2 D1; 5-FU 1,000 mg/m2 D1-4	40 Gy, 2 Gy/fr, 4 weeks	SCC	35/34	(8)
Le Prise et al, 1994	France	2 cycles: Cisplatin 100 mg/m2 D1,21; 5-FU 600 mg/m2 D2-5, 22–25	40 Gy, 20 Gy/10 fr, 2 weeks	SCC	41/45	(9)
Walsh et al, 1996	Ireland	2 cycles: Cisplatin 75 mg/m2 D1; 5-FU 15 mg/kg D1-5	40 Gy/15 fr, 3 weeks	AC	58/55	(10)
Bosset et al, 1997	France	2 cycles: Cisplatin 80 mg/m2 D1-2	37 Gy, 3.7 Gy/fr, 2 weeks	SCC	143/139	(11)
Urba et al, 2001	USA	2 cycles: Cisplatin 20 mg/m2 D1-5; 5-FU 300 mg/m2 D1-21; Vin 1 mg/m2 D1-4	45 Gy; 1.5 Gy/fr, 3 weeks	SCC 25%, AC 75%	50/50	(12)
An et al, 2003	China	2 cycles: 5-FU 1 mg/m2 D1-5, 21–25; cisplatin 25 mg/m2 D1, 22–25	36 Gy, 1.2 Gy/fr, 3 weeks	SCC	48/49	(13)
Lee et al, 2004	Korea	2 cycles: Cisplatin 60 mg/m2 D1; 5-FU 1,000 mg/m2 D3-5	45.6 Gy, 1.2 Gy/fr, 4 weeks	SCC 40%, AC 60%	51/50	(14)
Burmeister et al, 2005	Australia	One cycle: Cisplatin 80 mg/m2 D1; 5-FU 800 mg/m2 D1-4	35 Gy/15 fr, 3 weeks	SCC	128/128	(15)
Law et al, 2006	China	5-FU 500 mg/m2 D1-5, D24-28; cisplatin 100 mg/m2 D1,24	40 Gy, 2 Gy/fr, 3 weeks	SCC	170/109	(16)
Natsugoe et al, 2006	Japan	Cisplatin 7 mg over 2 h; 5-FU 350 mg over 24 h	40 Gy, 2 Gy/fr, 4 weeks	SCC	22/23	(17)
Cao et al, 2007	China	Cisplatin 20 mg/m2 D1-5, mitomycin 10 mg/m2/day D1, 5-FU 500 mg/m2 D1-5	40 Gy, 2 Gy/fr, 4 weeks	SCC	118/118	(18)
Jin et al, 2008	China	Cisplatin 20–30 mg/m2, Dl-5, 22–26; paclitaxel 135 mg/m2, D1,22	38–44 Gy, 2 Gy/fraction	SCC 90%, AC 10%	30/30	(19)
Peng et al, 2008	China	2 cycles: 5-FU 500 mg/m2 over 5 days, cisplatin 75 mg/m2 D1	40 Gy, 2 Gy/fraction 4 weeks	SCC	40/40	(20)
Tepper et al, 2008	USA	2 cycles: 5-FU 1,000 mg/m2 D1-4; cisplatin 100 mg/m2 D1	50.4 Gy, 1.8 Gy/fr, 5.6 weeks	SCC 25%, AC 75%	30/26	(21)
Lv et al, 2010	China	2 cycles: Cisplatin 20 mg/m2 D1-3, 22–25; paclitaxel 135 mg/m2 D1-3,22–25	40 Gy, 2 Gy/fr, 4 weeks	SCC	119/119	(22)
Jin et al, 2011	China	2 cycles: 5-FU 500 mg/m2 D1-5; cisplatin 75 mg/m2 D1	50 Gy, 2 Gy/fr, 5 weeks	SCC 90%, AC 10%	30/30	(23)
van Hagen et al, 2012	The Netherlands	5-week chemotherapy: Carboplatin AUC = 2, paclitaxel 50 mg/m2 D1, weekly	41.4 Gy, 1.8 Gy/fr, 4.6 weeks	SCC 23%, AC 75%LCC 2%	178/188	(24)
Yang et al, 2012	China	2 cycles: Cisplatin 75 mg/m2 D1,22; navelbine 25 mg/m2 D1,8,22,29	40 Gy, 2 Gy/fr, 4 weeks	SCC	54/59	(25)
Bass et al, 2014	Ireland	2 cycles: Cisplatin 75 mg/m2 D7; 5-FU 15 mg/kg D 1–5	40 Gy/15 fr, 3 weeks	AC	104/107	(26)
Mariette et al, 2014	France	2 cycles: Cisplatin 75 mg/m2 D1, 5-FU 800 mg/m2 on D1-4	45 Gy/25 fr, 5 weeks	SCC 72%, AC 28%	98/97	(27)
Shapiro et al, 2015	The Netherlands	5-week chemotherapy: Carboplatin AUC = 2, paclitaxel 50 mg/m2 D1, weekly	41.4 Gy, 1.8 Gy/fr, 4.6 weeks	SCC 23%, AC 75%	178/188	(28)

[i] SCC, squamous cell carcinoma; AC, adenocarcinoma; LCC, large-cell lung cancer; CRTS, chemoradiotherapy followed by surgery; SA, surgery alone; D, day; AUC, area under the curve; 5-FU, 5-fluorouracil; Vin, vinblastine.

Survival rate

The heterogeneity test at all the time points had a I2 value of <55%; thus, the fixed-effects model was used.

OSR1y, OSR3y and OSR5y outcomes of traditional and cumulative meta-analysis

Traditional meta-analysis provided evidence that, compared with the SA group, the OSR3y and OSR5y were significantly higher in the CRTS group. The pooled OSR3y was 44% (95% CI: 37–52%) vs. 30% (95% CI: 23–38%), respectively, and the OSR5y was 36% (95% CI: 32–42%) vs. 24% (95% CI: 19–29%), respectively, with an OR of 1.38 (1.20–1.58, P<0.001) and 1.42 (95% CI: 1.22–1.66, P<0.001), respectively. However, there was no statistically significant difference in OSR1y between the CRTS and SA groups; the pooled OSR1y was 71% (95% CI: 65–78%) vs. 68% (95% CI: 60–76%), respectively, and the OR was 1.06 (95% CI: 0.94–1.19, P=0.348) (Figs. 2A, 3A, 4A and 5; Table II).

Figure 2. Traditional and cumulative meta-analysis for 1-year overall survival rate. OR, odds ratio; CI, confidence interval.

Figure 3. Traditional and cumulative meta-analysis for 3-year overall survival rate. OR, odds ratio; CI, confidence interval.

Figure 4. Traditional and cumulative meta-analysis for 5-year overall survival rate. OR, odds ratio; CI, confidence interval.

Figure 5. Survival rate and (95% confidence interval) for CRTS and SA. OS1y, 1-year overall survival; OS3y, 3-year overall survival; OS5y, 5-year overall survival; CRTS, neoadjuvant chemoradiotherapy and surgery; SA, surgery alone.

Table II. Survival rate and surgical parameters of patients with EC by treatment approach.

Table II.

Survival rate and surgical parameters of patients with EC by treatment approach.

		No. of patients
Variables	No. of studies	CRTS	SA	OR (95% CI)	P-value
Survival rate
OSR1y	20	1,424	1,429	1.06 (0.94–1.19)	0.348
OSR3y	19	1,479	1,488	1.38 (1.20–1.58)	<0.0001
OSR5y	15	1,361	1,437	1.42 (1.22–1.66)	<0.0001
Surgery conditions
R0 resection rate	9	774	874	2.76 (2.15–3.53)	<0.0001
Local recurrence rate	10	668	679	0.49 (0.36–0.65)	<0.0001
Distant metastasis rate	10	668	679	0.76 (0.60–0.97)	0.02
Postoperative mortality	15	1,086	1,205	0.97 (0.72–1.32)	0.87

[i] The fixed-effects model was used. EC, esophageal carcinoma; CRTS, neoadjuvant chemoradiotherapy followed by surgery; SA, surgery alone; OR, odds ratio; CI, confidence interval; OSR, overall survival rate; y, year.

Cumulative meta-analyses were performed in chronological order. With the increase in the number of cases, OR point estimates and 95% CIs of all survival rates tended to be stable and exhibited an improving trend. When multiple studies with large sample sizes were added, the effect on the outcome was only a reduction in the length of the confidence interval, reflecting an increase in the accuracy of the estimated overall treatment response. Under the α=0.05 test standard, cumulative meta-analyses demonstrated there was no statistical difference between CRTS and SA in terms of OSR1y (Fig. 2B), and the P-value decreased gradually, stabilizing at P=0.334 (calculated via Microsoft Excel). As regards OSR3y (Fig. 3B), it was observed that the difference was initially confirmed to be statistically significant (OR=2.10, 95% CI: 1.18–3.72, P<0.05) when adding a 113 sample size study by Walsh et al (10) in 1996 under the selection criteria. The P-value was >0.05 when subsequent studies were added successively and the analysis was re-accumulated, and it again became <0.05 when including a 100 sample size study by Urba et al (12) in 2001 (OR=1.45, 95% CI: 1.04–2.02, P<0.05). Subsequently, the cumulative analysis of successively included studies demonstrated that the difference was statistically significant, with P-values stable at <0.05. As regards OSR5y (Fig. 4B), cumulative meta-analyses demonstrated that the difference was initially statistically significant in 2007, when a 102 sample size study was conducted by Cao et al (18) (OR=1.33, 95% CI: 1.06–1.66, P<0.05), after which time the P-values were stable at <0.05.

Surgical factors

The CRTS group had a significantly higher R0 resection rate and a lower local recurrence and distant metastasis rate compared with the SA group, with a pooled OR of 2.76 (95% CI: 2.15–3.53, P<0.001), 0.49 (95% CI: 0.36–6.65, P<0.001) and 0.76 (95% CI: 0.60–0.97, P=0.02), respectively; the differences were statistically significant. However, the incidence of postoperative mortality in the two groups suggested there was no significantly statistical difference, with an OR of 0.97 (95% CI: 0.72–1.32, P=0.87) (Fig. 6, Table II).

Figure 6. Traditional meta-analysis for surgical parameters of patients with EC by treatment schedule (CRTS or SA). CRTS, neoadjuvant chemoradiotherapy and surgery; SA, surgery alone; CI, confidence interval.

Subgroup analysis

Survival rate of squamous cell carcinoma and adenocarcinoma

The pooled OR of squamous cell carcinoma in terms of OSR3y and OSR5y in the CRTS and SA groups was 1.57 (95% CI: 1.21–2.04, P=0.0006) and 1.69 (95% CI: 1.32–2.16, P<0.0001), respectively; the differences were statistically significant. However, there was no statistically significant difference in terms of OSR1y (OR=1.13, 95% CI: 0.88–1.45, P=0.35). Compared with adenocarcinoma patients treated with SA, the OSR1y, OSR3y and OSR5y were significantly higher in CRTS, with an OR of 1.55 (95% CI: 1.09–2.20, P=0.01), 1.77 (95% CI: 1.34–2.36, P<0.0001) and 1.92 (95% CI: 1.34–2.75, P=0.0004), respectively; the differences were statistically significant (Table III).

Table III. Survival rate by histological type and continent in EC patients treated with CRTS and SA.

Table III.

Survival rate by histological type and continent in EC patients treated with CRTS and SA.

			No. of patients
Variables	Overall survival	No. of studies	CRTS	SA	OR (95% CI)	P-value
Histological type
SCC	OSR1y	11	647	654	1.13 (0.88–1.45)	0.35
	OSR3y	10	554	556	1.57 (1.21–2.04)	0.0006
	OSR5y	8	622	698	1.69 (1.32–2.16)	<0.0001
AC	OSR1y	4	295	302	1.55 (1.09–2.20)	0.01
	OSR3y	5	429	442	1.77 (1.34–2.36)	<0.0001
	OSR5y	4	371	387	1.92 (1.34–2.75)	0.0004
Location
Asia	OSR1y	8	398	403	1.05 (0.74–1.49)	0.80
	OSR3y	8	424	424	1.81 (1.37–2.40)	<0.0001
	OSR5y	7	452	514	1.73 (1.31–2.27)	<0.0001
Europe	OSR1y	7	669	673	1.22 (0.96–1.54)	0.10
	OSR3y	8	847	860	1.74 (1.42–2.14)	<0.0001
	OSR5y	5	701	719	1.69 (1.35–2.13)	<0.0001
USA	OSR1y	2	80	76	1.75 (0.86–3.55)	0.06
	OSR3y	2	80	76	3.55 (1.68–7.49)	0.0009
	OSR5y	2	80	76	2.80 (1.19–6.61)	0.02

[i] The fixed-effects model was used. EC, esophageal carcinoma; CRTS, neoadjuvant chemoradiotherapy and surgery; SA, surgery alone; OR, odds ratio; CI, confidence interval; OSR, overall survival rate; y, year; SCC, squamous cell carcinoma; AC, adenocarcinoma.

Survival rates of different countries or regions

The subgroup analysis of OSR3y, OSR5y for Asian, European and American populations were significantly higher in the CRTS group compared with those in the SA group, and the differences were all statistically significant (P<0.05). However, when comparing the OSR1y between the two groups in patients from the three continents, the difference was not significant (P>0.05; Table III).

Publication bias

A funnel plot analysis of all the studies was performed in the meta-analysis of OSR1y, OSR2y and OSR3y between CRTS and SA. This indicated that the publication bias was low in the present meta-analysis (Fig. 7).

Figure 7. Funnel plot for the publication bias tests. OSR1y, 1-year overall survival rate; OSR3y, 3-year overall survival rate; OSR5y, 5-year overall survival rate; OR, odds ratio; SE, standard error.

Discussion

CRT is quickly becoming the neoadjuvant treatment of choice for patients with resectable esophageal carcinoma prior to surgery. However, trials and meta-analyses on this subject are limited and varied, with small sample sizes and heterogeneity of population distribution characteristics, tumor pathological types, tumor location, radiation doses, chemotherapy regimens, surgical approach, postoperative care and adequacy of surgical resections, despite all the advantages of trimodality therapy.

In the CROSS trial (26), CRTS improved the long-term overall and progression-free survival in patients with resectable esophageal carcinoma; this improvement was statistically significant and clinically relevant for both the adenocarcinoma and squamous cell carcinoma subtypes. In addition, locoregional control and distant disease control also improved significantly. However, Mariette et al (25) reported that, compared with SA, CRTS with cisplatin plus fluorouracil did not improve R0 resection rate or survival, but rather enhanced postoperative mortality in patients with resectable esophageal carcinoma. Burmeister et al (15) obtained results in a randomised controlled phase III trial indicating that preoperative CRT with cisplatin and fluorouracil did not significantly improve progression-free or overall survival in patients with resectable esophageal cancer compared with SA.

Meta-analyses on CRTS vs. SA in esophageal cancer, however, are discordant. In the most recent meta-analysis of 13 studies on CRTS compared with SA in operable patients, the hazard ratio for all-cause mortality was 0.78 (P<0.001), favoring CRTS. However, due to the large majority of locally advanced cases included in the trials and the heterogeneity in staging methods, there was no definitive conclusion regarding survival benefit for stage I or II esophageal cancer (29). A meta-analysis of those trials by Gluud and Krag (30) reported a short-term survival benefit for neoadjuvant chemoradiotherapy over surgical monotherapy in adenocarcinoma as well as squamous cell carcinoma of the esophagus. In addition, a meta-analysis by Huang et al (31) reported that CRTS with paclitaxel plus platinum appeared to be a better choice compared with platinum plus 5-fluorouracil for esophageal cancer, particularly for squamous cell carcinoma. Wijnhoven et al (32) performed a secondary meta-analysis of six published meta-analyses to compare the differences in the studies included and statistical methods applied, and found heterogeneity between the RCTs included in the meta-analyses with regard to the previously mentioned content. Of note, the majority of RCTs were conducted in the 90s; hence, the diagnostic methods, staging, treatment delivery and outcome assessment reflected the clinical practice during tha decade.

Our aim was to conduct a meta-analysis combining the traditional and cumulative methods. The traditional meta-analysis revealed that CRTS may improve the long-term survival and surgical parameters, and reduce locoregional cancer recurrence and distant metastasis in adenocarcinoma as well as squamous cell carcinoma of the oesophagus, but there was no significant difference in terms of short-term survival. We focused more on the integration of various researches chronologically by using the cumulative meta-analysis. Clinical trials on a particular research topic constitute an increasing, open and continuous entity over time. Baum et al (33) first proposed the concept of cumulative meta-analysis that was first applied to clinical practice by Lau et al (34) on the basis of the traditional meta-analysis, adding studies sequentially and performing multiple meta-analyses in a sequential manner based on the time of publication, the size of the sample and the quality score of the study; whenever a new study is published, the meta-analysis may be again continued. Unlike traditional meta-analyses, which are performed only at a certain point in time, cumulative meta-analysis was performed at each time point in order to capture the variation tendency of the combined total effect, which may enable greater use of information, contribute to early detection of coherent interventions, and facilitate new research.

From forest plots of cumulative meta-analysis (performed in chronological order), it was observed that, as the number of cases increased, the test efficacy increased and the 95% CI gradually decreased; under the α=0.05 test standard, cumulative meta-analyses demonstrated there was no statistical difference between CRTS and SA in terms of OSR1y, and the P-value decreased gradually and stabilized at 0.334. Therefore, it was concluded that CRTS did not improve the short-term survival benefit of patients with esophageal cancer. The difference between the two treatment approaches in terms of OSR3y was initially confirmed to be statistically significant (OR=2.10, 95% CI: 1.18–3.72, P<0.05); when adding a 113 sample size study by Walsh et al (10) under the selected test criteria, it was observed that the treatment regimen was the same as that of previous studies, except that the subjects were adenocarcinoma patients rather than squamous cell carcinoma patients. Thus, it was hypothesized that CRTS may be more effective in treating esophageal adenocarcinoma. The same conclusion was reached using the traditional meta-analysis, as the OR for OSR1y, OSR3y and OSR5y in adenocarcinoma patients is higher compared with that in squamous cell carcinoma patients (1.55 vs. 1.13 for OSR1y, 1.77 vs. 1.57 for OSR3y and 1.92 vs. 1.69 for OSR5y). A meta-analysis conducted by Hai-Lin et al (35) also confirmed that CRTS may increase the survival rate of patients with esophageal adenocarcinoma. However, the P-value was >0.05 when adding a 282 sample size study by Bosset et al (11) in 1997, possibly due to the cisplatin monotherapy. The P-value again became <0.05 when a 100 sample size study by Urba et al (12) in 2001 was included (OR=1.45, 95% CI: 1.04–2.02, P<0.05), in which innovative triple therapy was used, combining vinblastine with cisplatin and fluorouracil. Liu et al (36) also reported in 2015 that cisplatin with vinorelbine may achieve a higher pathological complete response rate and better survival outcomes compared with cisplatin and fluorouracil in esophageal squamous cell carcinoma. Subsequently, the cumulative analysis of successively included studies demonstrated that the difference was statistically significant, with P-values stable at <0.05. It was demonstrated that CRTS may improve the 3-year survival benefit of patients with esophageal cancer. As regards OSR5y, cumulative meta-analyses demonstrated that the difference was initially found to be statistically significant in 2007, when a 102 sample size study was conducted by Cao et al (17) (OR=1.33, 95% CI: 1.06–1.66, P<0.05), after which the P-values were stable at <0.05. A study by Wolf et al (37) on long-term outcome of mitomycin C and 5-fluorouracil-based primary CRT for esophageal cancer demonstrated a significant increase of overall survival (P<0.0001) in the CRT vs. the radiotherapy alone group, indicating that CRTS may provide a long-term survival benefit to patients with esophageal cancer. However, it remains uncertain whether the alteration in the abovementioned treatment options is the cause of P<0.05, as this is only a monistic interpretation. From the present analysis, it was concluded that CRTS was able improve the long-term survival of patients with esophageal cancer, and may be more effective in treating esophageal adenocarcinoma. In addition, vinblastine or mitomycin combined with general chemotherapy were more likely to improve the long-term survival rate following complete resection, which may also be a future research focus.

Traditional meta-analysis may be associated with various types of bias, such as selection, implementation, exit and measurement bias; the same biases may occur at various time points in the cumulative meta-analysis and affect the determination of the overall effect trend. Furthermore, certain information could not be collected (e.g., the chronological cumulative effect of the treatment regimen, the difference in efficacy and the quality score of a single article), which is a major drawback. In addition, patients included in the present study were in various stages of the trial, such as adjuvant therapy; patient compliance was also different, which may affect the results. Furthermore, the 22 included studies differed significantly in sample size; thus, the contribution to the overall effect was not proportional, which was another limitation of the cumulative meta-analysis.

In summary, it may be concluded from the cumulative meta-analysis that CRTS may increase OSR3y and OSR5y by 38% (P<0.0001) and 42% (P<0.0001), respectively. From the forest plot, it was observed that the difference in OSR3y and OSR5y was statistically significant, with P-values stable at <0.05, indicating that CRTS may improve the patient survival rate. Therefore, it is recommended that the CRTS regimen is routinely used for patients with early resectable esophageal cancer. There are ongoing studies on this subject and, as the results of those studies are published, it may further elucidate the role of CRTS in the treatment of early resectable esophageal cancer.

References