Short-course versus Long-course Preoperative Radiotherapy plus Delayed Surgery in the Treatment of Rectal Cancer: a Meta-analysis
Background: Short-course preoperative radiation ( SCRT ) with delayed surgery has been shown to increase diseased complete response ( pCR ) rate in several tests. However, there was no clear reply on whether SCRT or long-course chemo-radiotherapy ( LCRT ) is more effectual. Therefore we conducted this meta-analysis to measure the safety and efficaciousness of SCRT with delayed surgery versus LCRT with delayed surgery for intervention of rectal malignant neoplastic disease.
Methods: Literatures were searched from PubMed, EMBASE, Web of scientific discipline, Cochrane Library and clinicaltrials.gov up to August, 2014. Quality of the randomized controlled tests ( RCTs ) was evaluated harmonizing to the Cochrane’s hazard of bias tool of RCT. The Grading of Recommendations Assessment, Development and Evaluation ( GRADE ) System was used to rate the degree of grounds. ReviewManager 5.3 was used for statistical analysis. Pooled hazard ratio ( RR ) and 95 % assurance interval ( CI ) were calculated.
Consequences: Four RCTs were included in the present systematic reappraisal. Three RCTs, with entire of 357 rectal malignant neoplastic disease patients, were included. Meta-analyses consequences demonstrated there were no significantly differences in sphincter saving rate, local return rate, grade 3~4 acute toxicity, R0 resection rate and downstaging rate. Compared with SCRT, LCRT significantly increased pCR rate [ RR=0.49, 95 % CI ( 0.31, 0.78 ) , P=0.003 ] .
Decisions: In footings of sphincter saving rate, local return rate, grade 3~4 acute toxicity, R0 resection rate and downstaging rate, SCRT with delayed surgery is every bit effectual as LCRT with delayed surgery for direction of rectal malignant neoplastic disease. LCRT significantly increased pCR rate compared with SCRT. Due to hazard of prejudice and impreciseness, farther multi-center big sample RCTs were needed to corroborate this decision.
Key words: rectal malignant neoplastic disease ; preoperative radiation therapy ; chemo-radiotherapy ; meta-analysis.
Long-course chemo-radiotherapy ( 45~50 Gy in 25 fractions ) with delayed surgery or short-course radiation therapy ( 25 Gy in 5 fractions ) with immediate surgery were the most frequent regimens for the intervention of localised and locally advanced resectable rectal malignant neoplastic disease ( 1-4 ) . Compared with postoperative chemo-radiotherapy, this scheme improved local control and was associated with decreased toxicity and better conformity of radiation therapy ( 2, 5, 6 ) . By comparing short-course preoperative radiation therapy with selective postoperative chemo-radiotherapy, the MRC ( Medical Research Council ) CR07 rectal test provided farther support for the short class regimen ( 5 ) . Long-course preoperative chemo-radiotherapy ( LCRT ) of 50.4 Gy in 6 hebdomads and 5 yearss with coincident chemotherapy has been widely used in the last decennaries. This regimen’s high quality, in footings of local control, was demonstrated in the German rectal malignant neoplastic disease test, when compared with postoperative chemo-radiotherapy ( 2, 7 ) . In fact, the optimum timing interval of surgery remains controversial. Short-course preoperative radiation with delayed surgery has been shown to increase diseased complete response ( pCR ) rate and bring on tumour down-staging rate in both randomized and experimental surveies in recent old ages ( 8-10 ) .
To day of the month, many surveies have explore whether SCRT or LCRT is more effectual neoadjuvant mode to better results for localised and locally advanced rectal malignant neoplastic disease [ 11-14 ] , but the consequences were heterogenous. ( 11-14 ) . On the one manus, the benefit of the short-course agenda is less expensive and more convenient, particularly in centres with a long waiting list or deficiency of medical resources ( 15 ) . On the other manus, long-course chemo-radiation might be better than the short-course irradiation agenda at increasing diseased complete response ( pCR ) and R0 resection rate ( 15 ) , because the tumour majority might be reduced before surgery. Therefore we performed this systematic reappraisal and meta-analysis to to the full analyse the safety and efficaciousness of SCRT with delayed surgery versus LCRT with delayed surgery as a mode for the direction of rectal malignant neoplastic disease.
Surveies would be included in systematic reappraisal if they met the undermentioned standards: ( 1 ) Participants: All the patients that were diagnosed as localised and locally advanced resectable rectal malignant neoplastic disease utilizing pathology and cytology were included in systematic reappraisal. Metastatic rectal malignant neoplastic disease patients were excluded. All the patients did non hold serious cardiorespiratory diseases and other terrible underlying diseases. ( 2 ) Interventions and comparings: Comparing the efficaciousness and safety of short-course radiation therapy with delayed surgery +/- accessory chemotherapy versus long-course radiation therapy or chemo-radiotherapy with delayed surgery +/- accessory chemotherapy for intervention of rectal malignant neoplastic disease. Short-course radiation therapy was non more than one hebdomad, and long-course radiation therapy should be more than 4 hebdomads at least. Entire dosage of short-course was at least more than 20 Gy, and long-course surpassed 45 Gy. Clocking interval of surgery was more than 4 hebdomads either SCRT or LCRT weaponries. ( 3 ) Results: The following results were evaluated: sphincter saving rate, R0 resection rate, Downstaging ( T phase decreased ) rate, pCR rate, local return rate and grade 3~4 acute toxicity. ( 4 ) Study design: randomized control tests ( RCTs ) .
We excluded the undermentioned articles: ( 1 ) The design of the survey was non randomized controlled tests, for illustration, retrospective survey, instance series or instance study etc. ; ( 2 ) The survey had repeated informations or did non describe results of involvement ; ( 3 ) Non-original research, such as reappraisal, remarks etc. ; ( 4 ) Short-course radiation therapy was more than one hebdomad, or long-course radiation therapy less than 4 hebdomads. Entire dosage of short-course was less than 20 Gy, or long-course less than 45 Gy. Clocking interval of surgery was less than 4 hebdomads either SCRT or LCRT weaponries.
Eligibility appraisal was performed independently in a non-blinded standardised mode by 2 referees. Any dissension between two writers was resolved by discussion..
We searched related articles in PubMed ( From 1966 to August 2014 ) , EMBASE ( From 1974 to August 2014 ) , Web of Science, Cochrane Library ( CENTRAL, Issue 8 of 12, Oct 2014 ) and clinicaltrials.gov up to August, 2014. In electronic searching, we used MeSH or Emtree footings combined free footings in all the hunt schemes. The undermentioned hunt footings were used: “rectal cancer” , “preoperative radiotherapy” , “chemoradiotherapy” , “neoadjuvant radiotherapy” , “short course” and “long course” . The whole hunt schemes were listed in the appendix. We besides reviewed the mentions of included surveies to look for potentially eligible articles. Furthermore, we checked abstracts that were published in major academic conferences ( American Society of Clinical Oncology, European Society for Medical Oncology, American Society for Therapeutic Radiology and Oncology and European SocieTy for Radiotherapy & A ; Oncology ) . No linguistic communication limitations were adopted.
Measuring hazard of prejudice of included RCTs
Qualities of included RCTs were evaluated harmonizing to the Cochrane Collaboration ‘s tool for measuring hazard of prejudice of RCT ( ROB tool, 5.1.0 ) ( 16 ) . The RoB tool included seven spheres: sequence coevals, allotment privacy, blinding of participants, forces and outcome assessors, uncomplete result informations, selective result coverage, and other beginnings of prejudice. For each survey, we made judgements about hazard of prejudice from each of the seven spheres of the tool. In all spheres, an reply ‘Yes’ indicated a low hazard of prejudice, an reply ‘No’ indicated high hazard of prejudice, and if deficient item is reported of what happened in the survey, the judgement would normally be ‘Unclear’ hazard of prejudice.
Data extraction was performed wholly independently by two referees. Reviewers were non blinded to writers or diaries. Disagreements were resolved by treatment between the two reappraisal writers ; if no consensus was reached, a 3rd writer would make up one’s mind. Information about baseline patient features, continuance of follow up, and the figure of events for all the results, test design, intercessions and results were extracted from each included survey.
All statistical analyses were performed utilizing ReviewManager 5.3 package. Hazard ratio ( RR ) and 95 % assurance intervals ( CI ) was calculated for count informations. Chi-square trial and I-square trial were used for proving heterogeneousness between surveies. If heterogeneousness was non present ( P & gt ; 0.10, I2& lt ; 50 % ) , fixed-effect theoretical account would be adopted for analysis, otherwise, random-effect would be employed. In the presence of heterogeneousness, we explored possible beginnings from the undermentioned three facets: clinical, methodological and statistical. In the instance of inordinate heterogeneousness, descriptive analysis instead than meta-analysis was employed.
Quality of grounds
The Grading of Recommendations Assessment, Development, and Evaluation ( GRADE ) attack was used to rate the quality of grounds and strength of recommendations ( 17-20 ) . Hazard of prejudice, restrictions, the indirectness, the consistence of the consequences across surveies, the preciseness of the overall estimation across surveies and other considerations are six spheres of the tool. For each result, if farther research was really improbable to alter our assurance in the estimation of consequence, the quality of the grounds was rated as high ; if farther research was likely to hold an of import impact on our assurance in the estimation of consequence and may alter the estimation, the quality was moderate ; if farther research was really likely to hold an of import impact on our assurance in the estimation of consequence and is likely to alter the estimation, the quality was low ; if the estimation was really unsure, the quality was really low. The GRADEpro 3.6 package was used to gauge the quality of the grounds in the meta-analysis by two referees. If there were dissensions between the two referees, a 3rd writer would fall in to discourse and do determinations.
Study choice and features of included surveies
Entire 502 relevant literatures were searched, 165 extras were removed. After reviewed the rubrics and abstracts of 337 of records, 330 of them was excluded due to evidently irrelevancy. 7 of full-text were obtained to further find eligibility. We ruled out another 3 of full-text articles: 1 article due to intercessions non run intoing inclusion standards ; 1 article due to reexamine ; 1 article due to non-RCT. Finally, 4 surveies were included in the qualitative systematic reappraisal ( 10, 21-23 ) , 1 of them is an on-going test, no publications were reported ( 23 ) ; 3 tests, entire 357 patients, were included in meta-analysis ( 10, 21, 22 ) . The PRISMA flow diagram of surveies was shown in Figure 1. The features of the surveies were shown in Table 1.
Hazard of bias appraisal
This meta-analysis included 3 RCTs: the baseline features of patients were reported in all RCTs. All 3 RCTs mentioned “ random ” ; merely 1 RCT reported an equal randomised sequence coevals and allotment privacy ( 10 ) , 1 RCT was ill-defined ( 21 ) , 1 RCT was with high hazard ( 22 ) . 2 RCTs described the grounds of uncomplete result informations ( 10, 21 ) , 1 RCT with high hazard ( 22 ) . All tests did non advert whether the blind method was adopted or non, nevertheless, this should unlikely impact the quality appraisal consequences ( Figure 2 ) .
Consequences of meta-analysis
Sphincter saving rate
Two tests reported the sphincter saving rate ; entire 172 rectal malignant neoplastic disease patients were included in the meta-analysis. There was no important difference in the sphincter saving rate between SCRT and LCRT weaponries [ RR=1.14, 95 % CI ( 0.86, 1.52 ) , p=0.37 ] . No obvious heterogeneousness was found ( I2=24 % , P=0.25 ) , so the fixed consequence theoretical account was employed, ( Figure 3A ) .
R0 resection rate
Two tests, with entire 127 rectal malignant neoplastic disease patients, were included in the meta-analysis to measure R0 resection rate. Meta-analysis suggested that there was no important difference between SCRT and LCRT weaponries [ RR=0.94, 95 % CI ( 0.84, 1.06 ) , p=0.29 ] , without important heterogeneousness was detected ( I2=0 % , P=0.81 ) , so the fixed consequence theoretical account was applicable, ( Figure 3B ) .
Downstaging ( T phase decreased ) rate
Merely one RCT measure the downstaging rate. Compared SCRT with LCRT, there was no important difference between the two weaponries [ RR=0.62, 95 % CI ( 0.33, 1.16 ) , P=0.14 ] , ( Figure 3C ) .
Two RCTs reported the pCR rate ; entire 172 rectal malignant neoplastic disease patients were included in the meta-analysis. LCRT significantly increased the pCR rate compared with SCRT [ RR=0.49, 95 % CI ( 0.31, 0.78 ) , P=0.003 ] , without heterogeneousness was found ( I2=0 % , p=0.32 ) , so fixed consequence theoretical account was adopted, ( Figure 4A ) .
Local return rate ( LRR )
Merely one RCT reported the local return rate. There was no important difference between SCRT and LCRT weaponries [ RR=4.76, 95 % CI ( 0.66, 34.53 ) , p=0.12 ] , ( Figure 4B ) .
Grade 3~4 acute toxicity
Two RCTs reported the class 3~4 acute toxicity ; entire 274 rectal malignant neoplastic disease patients were included in the meta-analysis. There was no important difference between SCRT and LCRT weaponries [ RR=0.60, 95 % CI ( 0.19, 1.88 ) , P=0.38 ] , without obvious heterogeneousness was detected ( I2=17 % , p=0.27 ) , so the fixed consequence theoretical account was employed, ( Figure 4C ) .
Quality of grounds
There were 6 results in this meta-analysis. Sphincter saving rate, pCR rate, local return rate and grade 3~4 acute toxicity were critical results ; R0 resection rate and downstaging rate were both of import results. The quality of the grounds of each result was shown in table 2.
This systematic reappraisal and meta-analysis manifested that there were no evidently differences in sphincter saving rate, local return rate, grade 3~4 acute toxicity, R0 resection rate and downstaging rate between two weaponries. Compared with SCRT plus delayed surgery, LCRT with delayed surgery evidently increased pCR rate [ RR=0.49, 95 % CI ( 0.31, 0.78 ) , P=0.003 ] . Based on the GRADE system, the grounds qualities of pCR rate and R0 resection rate were “moderate” ; the grounds qualities of sphincter saving rate, downstaging rate and grade 3~4 acute toxicity were “low” ; local return rate was “very low” . The chief ground was hazards of prejudice and impreciseness.
Overall completeness and pertinence of grounds
Most of rectal malignant neoplastic disease patients included in this meta-analysis were clinical phase II~III and without distant metastasis. SCRT with delayed surgery would be a good pick for T3~4 and/or N+ rectal malignant neoplastic disease patients. So the consequences of this meta-analysis could be applicable to the patients with localised and locally advanced rectal malignant neoplastic disease patients ( phase II~III, without distant metastasis ) . Based on the major findings of this systematic reappraisal, there were no important differences in most of results between the two intercessions. The benefit of the SCRT is more convenient and cheap, particularly in centres with a long waiting list or deficiency of medical resources ( 15 ) . Furthermore, LCRT is better than SCRT at increasing diseased complete response rate. Therefore, based on the available grounds, take which sort of intervention scheme mostly depends on the clinician’s experience, the patient ‘s clinical characteristics and the public wellness resources ( 14, 24, 25 ) .
Merely 3 RCTs, entire 357 rectal malignant neoplastic disease patients, were included in the meta-analysis, so the sample size was excessively little to observe the possible statistical difference in some results, such as sphincter saving rate, downstaging rate, local return rate and grade 3~4 acute toxicity, which were with serious or really serious impreciseness ( 19 ) . Take grade 3~4 acute toxicity as illustration, the statistical power of the meta-analysis was merely 0.2881, doubtless, it is non plenty ( 26 ) . In add-on, there were some possible hazards of prejudices. First, though all included RCTs mentioned “ random ” , merely 1 RCT reported the method of equal randomized sequence coevals ( 10 ) , 1 RCTs was ill-defined ( 21 ) , 1 RCT with high hazard ( 22 ) . Second, merely 1 RCT reported the inside informations about the method of allotment privacy ( 10 ) , 1 RCT was ill-defined ( 21 ) , 1 RCT with high hazard ( 22 ) . So selection prejudices and abrasion prejudices were inevitable ( 16 ) .
It should be noted that clinical heterogeneousness existed in single surveies. There were some possible beginnings of clinical heterogeneousness. First, though most recruited rectal malignant neoplastic disease patients with clinical phase II~III, there was still some phase I and IV patients, and that the proportion of different clinical phase of the patients included in each survey was non wholly comparable, which might be an of import beginning of heterogeneousness. Second, the type and dosage of the same chemotherapeutics, every bit good as path of disposal in each survey, was non comparable, which may present heterogeneousness. Third, TME was performed in most surveies, except the survey performed by Bujko et al 2013, in which local deletion was adopted ( 22 ) . Last but non least, continuance of followup was different in each survey. In one word, these restrictions might take to possible prejudices in the systematic reappraisal procedure. In add-on, all included RCTs did non describe long-run follow up consequences, such as overall endurance, patterned advance free endurance and distant metastases return rate. So we could non objectively and comprehensively evaluation the efficaciousness of two intervention schemes and farther surveies were needed to turn to this issue.
In decision, SCRT with delayed surgery is every bit effectual as LCRT with delayed surgery for intervention of rectal malignant neoplastic disease in footings of sphincter saving rate, local return rate, grade 3~4 acute toxicity, R0 resection rate and downstaging rate. LCRT significantly increased pCR rate compared with SCRT. Due to serious hazard of prejudice and impreciseness, this decision necessitate farther multi-center big sample RCTs with longer-term follow up to corroborate it.
Table 1. Features of tests included in systematic reappraisal
Table 2. GRADE grounds profile
Figure 1.Flow chart of the survey choice procedure
Figure 2.( A ) Hazard of bias graph: reappraisal writers ‘ judgements about each hazard of bias point presented as per centums across all included surveies ; ( B ) Hazard of bias drumhead: reappraisal writers ‘ judgements about each hazard of bias point for each included survey.
Figure 3.Meta-analysis of short-course RT plus delayed surgery vs. long-course chemo-RT plus delayed surgery. ( A ) : Meta-analysis consequences of Sphincter saving rate ; ( B ) : Meta-analysis consequences of R0 resection rate ; ( C ) : Meta-analysis consequences of downstaging rate.
Figure 4.Meta-analysis of short-course RT plus delayed surgery vs. long-course chemo-RT plus delayed surgery. ( A ) : Meta-analysis consequences of pCR rate ; ( B ) : Meta-analysis consequences of Local return rate ( LRR ) ; ( C ) : Meta-analysis consequences of Grade 3~4 ague toxicity.