The Effectiveness of Intensity Modulated Radiation Therapy versus Three-Dimensional Radiation Therapy in Prostate Cancer: A Meta-Analysis of the Literatures

Background and Purpose Intensity modulated radiation therapy (IMRT) can deliver higher doses with less damage of healthy tissues compared with three-dimensional radiation therapy (3DCRT). However, for the scenarios with better clinical outcomes for IMRT than 3DCRT in prostate cancer, the results remain ambiguous. We performed a meta-analysis to assess whether IMRT can provide better clinical outcomes in comparison with 3DCRT in patients diagnosed with prostate cancer. Materials and Methods We conducted a meta-analysis of 23 studies (n = 9556) comparing the clinical outcomes, including gastrointestinal (GI) toxicity, genitourinary (GU) toxicity, biochemical controland overall survival (OS). Results IMRT was significantly associated with decreased 2–4 grade acute GI toxicity [risk ratio (RR) = 0.59 (95% confidence interval (CI), 0.44, 0.78)], late GI toxicity [RR = 0.54, 95%CI (0.38, 0.78)], late rectal bleeding [RR = 0.48, 95%CI (0.27, 0.85)], and achieved better biochemical control[RR = 1.17, 95%CI (1.08, 1.27)] in comparison with 3DCRT. IMRT and 3DCRT remain the same in regard of grade 2–4 acute rectal toxicity [RR = 1.03, 95%CI (0.45, 2.36)], late GU toxicity [RR = 1.03, 95%CI (0.82, 1.30)] and overall survival [RR = 1.07, 95%CI (0.96, 1.19)], while IMRT slightly increased the morbidity of grade 2–4 acute GU toxicity [RR = 1.08, 95%CI (1.00, 1.17)]. Conclusions Although some bias cannot be ignored, IMRT appears to be a better choice for the treatment of prostate cancer when compared with 3DCRT.


Introduction
Prostate cancer ranks the most common cancer and the second most common cause of cancer death in men [1]. Radiation therapy (RT) is widely used in the treatment of prostate cancer [2][3][4][5][6]. Dose escalation has been generally adopted in the RT of prostate cancer for its advantage of improved tumor control outcomes [7][8][9][10][11][12][13][14]. Since most of the patients who were diagnosed with non-metastatic prostate cancer can survive longer than 10 years, the choice of RT techniques with minimized RT-related toxicity is important for improving quality of life [15][16][17][18][19]. However, higher doses are linked to increased normal tissue toxicity, such as late gastrointestinal (GI) toxicity and late genitourinary (GU) toxicity [7,20].
As technology advances, new RT techniques have emerged and have been used in clinical practice. Three-dimensional conformal radiation therapy (3DCRT) delivers a radiation dose conforming to the target volume of tumor [21]. Thus 3DCRT significantly increases the target dose whilereducing the exposure of healthy tissue [2,21,22]. RT techniques evolved to an advanced form of 3DCRT, intensity modulated radiation therapy (IMRT), which generates non-uniform fields to increase the radiation dose delivered to the intended target volume while potentially minimizing the irradiation to the organs at risk [23,24]. Nevertheless, the probability of a marginal miss is a potential weakness of IMRT. Besides, the dosehomogeneity, increase of irradiation doses to larger volumes of healthy tissuesand longer time required for planning need to be considered in the application of IMRT [25,26]. The increased total body exposure and monitor units raise the risk of second malignancies of IMRT in comparison with conventional RT [27][28][29][30].
However, the potential benefits of IMRT over 3DCRT for prostate cancer treatment have not yet been clarified. Therefore, this meta-analysis was conducted to assess whether IMRT could improve clinical outcomes in comparison with 3DCRT in patients diagnosed with prostate cancer, including acute GI toxicity, acute GU toxicity, acute rectal toxicity, late GI toxicity, late GU toxicity, late rectal bleeding, biochemical controland overall survival (OS).

Primary search strategy
The PubMed (MEDLINE) and EMBASE were searched for relevant publications by combining search terms "prostate cancer [Title/ Abstract]", "Intensity modulated radiation therapy [Title/ Abstract]", "IMRT[Title/ Abstract]", "Three dimensional conformal radiation therapy [Title/ Abstract]", and "3DCRT [Title/ Abstract]". There was no date of publication limits and the most recentliterature was published on July 25 th , 2015. Only studies in English were included. Furthermore, reference lists from primarily identified studies were also manually searched.

Criteria for considering studies in this review
Eligible studies had to compare IMRT with 3DCRT in patients diagnosed with prostate cancer. Those studies were then selected according to the following criteria: (1) Studies with GI, GU toxicity or other clinical outcomes, including RFS or OS, were included in this meta-analysis.
(2) Late GI and late GU toxicity were scored according to the Fox Chase (FC) modification of the Radiation Therapy Oncology Group (RTOG) and Late Effects Normal Tissue Task Force (LENT) toxicity criteria (RTOG/FC-LENT late toxicity criteria)/Common Terminology Criteria (CTC) (version 2.0, 3.0 or 4.0) [31]. (3) Late rectal bleeding was scored based on RTOG criteria [32]. (4) Biochemical failure was defined as a rise in prostate-specific antigen (PSA) level of 2 ng/ml above the nadir, with no backdating (ASTROPhoenix definition) [33]. Two reviewers conducted a primary assessment independently to confirm the eligibility of the abstracts searched from database. Discrepancies were solved by cooperative discussion. The names of all authors and medical centers involved in each study were carefully examined in order to avoid duplicated data. If duplicated studies were found, the studies with the largest number of patients were retained.

Data Extraction
Data were carefully extracted independently from all the included publications by two reviewers, using a standardized data collection form. Data extraction included the following items: author, year, study design, sample size, planning target volume (PTV), total dose of RT, fraction dose, margin, method for dose prescription, image guidance, tumor stage, median followup time, percentage of androgen deprivation therapy (ADT), andscore criteria.

Statistical Analysis
Included publications were divided into eight groups for analysis: those with data regarding acute GI toxicity, acute GU toxicity, acute rectal toxicity, late GI toxicity, late GU toxicity, late rectal bleeding, biochemical control, and OS.
For the quantitative aggregation of outcomes, the impacts of treatment on acute toxicity, late toxicity, RFS and OS of each publication were reported for by estimating RRs with 95% confidence interval (CI) value. The RR and its 95% CI were extracted from the original article. If RR and its 95% CI were not available, the total number of events and number of patients at risk in each group were extracted to estimate RR and its 95% CI, according to the methods described by Parmer et al [34]. Eventually, Kaplan-Meier curves were read using Engauge Digitizer version 4.1 (free software downloaded from http://sourceforge.net) to extract data to reconstruct RR and its 95% CIwhen the exploitable data were only presented in the form of figure.
To assess the heterogeneity of the publications, a fixed effect model was used for meta-analysis. If the I 2 was higher than 50%, a random effect model was used. Conventionally, the difference would be considered statistically significant if the 95% CI of RR did not overlap the value 1 with p < 0.05. Study estimates, together with pooled estimates, were presented in the form of forest plots. Publication bias was assessed graphically by funnel plots and Egger's linear regression method was used to assess the funnel plot asymmetry. (p < 0.05 was considered to be statistically significant) [35]. The meta-analysis was done with Stata version 12.0 (Stata Corporation, College Station, TX, USA).

Study selection and characteristics
The initial search algorithm retrieved 2656 references and 146 candidate studies were fully evaluated. Upon further review, 23 articles met the eligibility criteria, and the other 123 articles were out of scope. The flowchart of the literature search is shown in Fig 1. The total number of the included patients was 9556, ranging from 27 to 1571 per study. The main characteristics of included studies are presented in Table 1. The study design was more often a retrospective (n = 16) than a prospective cohort study (n = 5). The prescribed doses to the primary tumor were 70-85.3 Gy in IMRT group and 55.8-84.8 Gy in 3DCRT group. Stage I/II comprised 77.3% of the patients, and the remaining 22.7% were in stage III/IV. The median follow-up time ranged from 5.3 months to 120 months.
Of the included studies, 14 studies compared the effects of acute toxicity of an IMRT group to that of a 3DCRT group, including acute GI toxicity (n = 12), acute GU toxicity (n = 12) and acute rectal toxicity (n = 4). Additionally, 21 studies compared the late toxicity effects of IMRTgroup to that of 3DCRT group, including late GI toxicity (n = 13), late GU toxicity (n = 12) and late rectal bleeding (n = 5). Furthermore, 6 studies compared the biochemical controlbetween IMRT group and 3DCRT group, and 3 studies compared the OS between IMRT group and 3DCRT group ( Table 2).   (Fig 2A). Due to obvious heterogeneity, random effect model was employed.

Overall survival
Data regarding overall survival were available in three studies with 924 patients [37,41,44]. A non-significant increase in overall survival favoring IMRT was found [RR = 1.07, 95%CI (0.96, 1.19)] (Fig 3B). Random effect model was performed for the obvious heterogeneity.

Publication bias
Both Begg's funnel plot and Egger's test were employed to assess the publication bias in all studies evaluating acute GI toxicity, acute GU toxicity, acute rectal toxicity, late GI toxicity, late GU toxicity, late rectal bleeding, biochemical control, and OS, respectively (Fig 4). The Begg's funnel plot did not indicate any evidence of statistically significant asymmetry in the metaanalysis of acute GI toxicity (p = 0.784), acute GU toxicity (p = 0.661), acute rectal toxicity

Discussion
In this meta-analysis, we enrolled 23 eligible studies comparing the clinical outcomes between IMRT and 3DCRT in patients diagnosed with prostate cancer. The present study showed that IMRT was associated with decreased 2-4 grade acute GI toxicity, late GI toxicity, and late rectal bleeding compared with 3DCRT. However, IMRT significantly increased grade 2-4 acute GU toxicity with similar grade 2-4 late GU toxicity. Moreover, no significant differences were discovered in grade 2-4 acute rectal toxicity and overall survival. Nevertheless, IMRT showed improved biochemical control than 3DCRT,suggesting better PSA relapse-free survival in IMRT. These results imply that IMRT might be superior to 3DCRT with less toxicity and better PSA relapse-free survival in patients diagnosed with prostate cancer. However, more high quality studies will be needed to further identify this result. Compared with 3DCRT, IMRT can deliver radiation with the capability of intensely conforming to cancerous site, which means IMRT can deliver higher dose to the target volume with less damage of normal tissues and with the creation of steep dose gradients and concave dose distribution [59,60]. On the one hand, dose-escalated RT has been demonstrated to generate better biochemical control when compared with lower doses by some randomized trials [8,61]. On the other hand, higher doses were associated with increased RT related side effects. Therefore, IMRT is generally believed to minimize treatment related toxicity and relatively improve survival. Besides, IMRT can also be performed to increase the homogeneity of dose distribution [59].
IMRT also has some drawbacks. Compared with 3DCRT, IMRT leads to larger volumes of healthy tissues exposed to low doses of radiation, which may increase the risk of second malignancies. However, more solid data are needed to clarify the clinical relevance [27,62]. Furthermore, IMRT is a kind of complex RT technique, which needs longer delivery time and has higher requirements for the physicists [63]. IMRT is estimated to cost about £1100 more than 3DCRT, which mainly comes from additional radiographer, medical and physics staff time  [64]. Nevertheless, there is still a need to understand the cost-effectiveness of IMRT, which may produce more quality-adjusted life-years (QALYs) with lower total costs [64]. Hence, it is important to assess the benefits and risks of IMRT.
In the published trials of RT, GI and GU toxicities are the most frequently studied and may deeply influence quality of life in patients who are diagnosed with prostate cancer [65][66][67][68][69]. Rectal bleeding is a type of late GI toxicity, but it sometimes is reported as a sole end point due to its objectivity [70][71][72][73]. Therefore, in this meta-analysis, we assessed not only the PSA relapse free survival and overall survival, but also the GI and GU toxicity and late rectal bleeding between 3DCRT and IMRT. However, randomized controlled trials that compare the clinical efficacy of IMRT with 3DCRT are still lacking.
Although meta-analysis is considered the gold standard by some authors, some potential bias cannot be completely eliminated. Begg's funnel and Egger's test were employed in this meta-analysis, and no statistically significant publication bias was discovered. However, several aspects which may produce potential biases in this meta-analysis should be discussed. First, only the literatures published in English were included because of the inaccessibility of other languages for reviewers. So the literature published in other languages, such as German, French and Spanish, was excluded in this meta-analysis [74][75][76]. This selection may cause further approval of the positive results, because positive results usually are published in English, while negative results tend to be published in native languages. This is called "file-drawer problem". Second, some studies were excluded due to the inaccessibility of extractingestimated RR value. One example of this is a study that compares the toxicity between 3DCRT and IMRT in the treatment of localized prostate cancer. In this study, they found a significant difference in late GU morbidity between 3DCRT and IMRT (p = 0.025).However, no data was available about late GU toxicity for meta-analysis from this study [58].
Third, the obvious heterogeneity between studies may be derived from different characteristics of study design, including different sample size, tumor stage, combined therapy, previous treatments, follow-up time, dose of the radiation therapy, etc. For example, two included studies reported that all of their patients had a prostatectomy, while only 15.9% to 54.9% of the patients in the remaining studies had a prostatectomy before radiotherapy [37]. Besides, the doses of the radiation varied among studies. Most of the studies used prescribed doses of 70 to 78 Gy, while one study performed a median dose up to 85.3 Gy, which may produce a different effect on the morbidity of GU or GI toxicity [58].
One limitation of this study is that we ignored the effect of combination treatments, and were not able to stratify patients according to whether they received surgery. In those studies which contain patients who underwent surgery, only one study analyzed the relationship between surgery and the incidence of late GU toxicity [40,41,45,46,51,56]. In this study, the actuarial 5-year risk of late GU toxicity was significantly higher in patients with a history of prostatectomy than in those without surgery [HR = 2.35 (95%CI 1.17-4.71)] [46].The other studies only compared the clinical outcomes of 3DCRT with IMRT without analyzing the influence of surgery on different RT technologies. Therefore, based on the insufficient data, we can't analyze the effect of surgery, on survival or toxicity. As for hormone therapy, only 3 studies discussed the influence of hormone therapy on late toxicity [46,47,50]. One study analyzed the influence of hormone therapy on late rectal toxicity, and concluded that hormone therapy had no significant influence on the risk of late rectal toxicity [HR = 2.59, 95% CI (1.00, 6.70), p = 0.10] [47]. Data regarding late GU toxicity was available in the other two studies [46,50]. Subgroup analysis showed no significant influence of hormone therapy on incidence of late GI and late GU toxicity ([HR = 0.47, 95% CI (0.16, 1.39)], [HR = 0.65, 95% CI (0.42, 1.01)], respectively) (S1 Fig). Therefore, we concluded that hormone therapy might have no influence on the occurrence of late GI or GU toxicity in the treatment of 3DCRT and IMRT. Another limitation is that we did not stratify the patients according to recurrence risk. Only two studies separated their patients into low, intermediate and high risk groups, which was not enough for us to perform a subgroups analysis with such small numbers of studies [41,44]. More researches investigating the associations of risk group and radiotherapy are needed.
In conclusion, IMRT significantly decreases the occurrence of 2-4 grade acute GI toxicity, late GI toxicity, late rectal bleeding, and achieves better PSA relapse free survival in comparison with 3DCRT. IMRT and 3DCRT remain the same in regard of acute rectal toxicity, late GU toxicity and overall survival, while IMRT increases the morbidity of acute GU toxicity. In general, based on the above results, IMRT should be considered as a better choice for the treatment of prostate cancer. More randomized controlled trials are needed to determine the subset of patients diagnosed with prostate cancer.