Genitourinary and gastrointestinal toxicity among patients with localized prostate cancer treated with conventional versus moderately hypofractionated radiation therapy: systematic review and meta-analysis

Abstract

Background: Hypofractionated (HRT) prostate radiation therapy has the potential to deliver a higher biologically effective dose over a shorter time compared with conventional fractionation (CRT). HRT, giving fewer fractions each with higher dose, might improve the therapeutic ratio, resource use and patient convenience but the toxicity is still controversial. Our objective was to compare the gastroinstestinal (GI) and genitourinary (GU) toxicity of HRT versus CRT.

Methods: Systematic review and meta-analysis of randomized clinical trials studies in PubMed, Cochrane and EMBASE databases published through December 2016 was done. Only randomized trials that evaluated patients with localized prostate cancer (PCa) undergoing CRT or HRT were included. In these studies, the daily dose was 1.8 Gy or 2 Gy per day for CRT and 2.4 to 3.4 Gy for HRT.

Results: 7317 patients in nine studies were analyzed. Six studies included acute GU toxicity data which showed similar rates for both HRT and CRT (32.6vs. 31.9%; RD 0.00; 95% CI; −0.03,0.03; p = .81; I² = 0%). Similarly, seven studies showed no difference in late GU toxicity based on treatment schedule (28.7 vs. 28.0%; RD −0.01; 95% CI; −0.04,0.03; p = .67; I² = 52%). GI toxicity at three months after radiotherapy was higher in patients treated with HRT in six studies (27.5 vs. 21.9%; RD 0.06; 95% CI; 0.02,0.10; p = .004; I² = 39%); however, eight studies showed GI toxicity 12 months or more after radiotherapy that was statistically the same (12.9 HRT vs. 16.2% CRT; RD −0.01; 95% CI; −0.04,0.02; p = .41; I² = 58%).

Conclusion: In meta-analysis of the available randomized trials on moderate HRT versus CRT for prostate cancer, acute and late GU toxicity were similar for both treatment schemes. While HRT was associated with higher acute GI toxicity, late toxicity was similar.

Introduction

Multiple studies, including randomized clinical trials, have shown that increasing radiation dose for prostate cancer improves local tumor control and decreases metastasis [1]. Additionally, associated complication rates have been proven to be acceptable when advanced techniques such as intensity-modulated radiation therapy (IMRT) and image-guidance are used. This work was first done with conventional radiotherapy fractionation (CRT) and doses up to 79 Gy was used.

Based on evidence that prostate cancer has a low alpha/beta ratio, larger fraction size was thought to be a way to exploit the differential between tumor and normal tissue [2]. Reports on the use of hypofractionated radiotherapy (HRT) schemes began to be published in the 1990s from countries with socialized medicine and in which treatment centers were sometimes distant from patients, such as the United Kingdom, Canada and Australia [3]. Using moderately hypofractionated radiation schemes (less than 5 Gy/day) could also be beneficial in lowering treatment costs, decreasing the occupancy of linear accelerators leading to decreased waiting times for therapy and in the patient convenience of fewer fractions.

HRT for prostate cancer has the potential to deliver a higher biologically effective dose (BED) over a shorter time compared with CRT [4]. However, the concern is the toxicity to the nearby rectum and bladder with the higher BED. Several randomized clinical trials reporting gastrointestinal (GI) and genitourinary (GU) toxicity have now been published. This is a systematic review and meta-analysis of the contemporary randomized clinical trials comparing moderate HRT to CRT focusing on toxicity outcomes.

Material and methods

Literature search

The search was performed using PubMed, Cochrane and Scielo for publications up to April 2017. Randomized trials published in English, Spanish or Portuguese were identified through a computerized blinded search using two sensitive search strategies combining the following medical subject headings and keywords: “((Radiotherapy or radiation therapy) and (prostate cancer OR prostatic neoplasm)). These were filtered by the “Titles/Abstract” category. All reviewed articles and cross-referenced studies from the retrieved articles were screened for pertinent information.

Inclusion and exclusion criteria

As seen in Table 1, we defined inclusion criteria for the literature search using the ‘Population, Intervention, Control, Outcome, Study Design (PICOS)’ approach.

Table 1. Studies characteristics.

					Gleason			Hypofractinated Protocol	Conventional protocol		ADT
Article	Year	Type of study	Median age	PSA	score	TNM		(Gy/fractions/weeks)	(Gy/fractions/weeks)	Pelvic RDT	HRT	CRT	Follow-up	IS	FS
Pollack et al. [5,6]	2006, 2013	RCT	Interval	10–80	>5	T1-3N0M0		64.6/19/6.5	78/39/8	Y	44%	47%	68.4 months	307	303
Marzi [7]	2009	RCT	Interval	>10	≥7	T2c-T4N0M0		63/20/5	76/38/8	N	100%	100%	30 months	162	114
Hoffman et al. [8]	2013	RCT	69 × 67	≤20	<10	T1b-T3bN0M0		72/30/NA	75.6/42/NA	N	25%	23%	60 months	222	203
Norkus et al. [9]	2013	RCT	64 × 66	>20	8 to 10	T3a-3bN0M0		62/20/5	80/40/8	Y	100%	100%	3 months	124	112
Dearnaley et al. [12,13]	2012, 2016	RCT	68 × 67 × 68	<30	<8	T1b-3aN0M0		60/20/4 and 57/19/3.8	74/37/7.4	N	100%	100%	60 months	3216	3133
Aluwini et al. [15,16]	2015, 2016	RCT	70 × 71	<60	NA	T1b-T4N0M0		70.2/26/NA	76/38/NA	N	66%	67%	60 months	820	782
Lee et al. [17]	2016	RCT	interval	<10	6	T1b-T2cN0M0		70/28/5.6	73.8/41/8.2	N	0%	0%	69.6 months	1092	1030
Catton et al. [18]	2017	RCT	72 × 71	≤20	≤7	T1-2cN0M0		60/20/4	78/39/7-8	N	6%	6%	72 months	1206	1192
Arcangeli [10,12]	2010, 2017	RCT	75 × 75	≤100	NA	T1-T3bN0M0		62/20/5	80/40/8	N	100%	100%	35/32 months	168	168
									BED CRT		BED HRT			EQD2 HRT
Article	Year	GU Toxicity	GI toxicity	Alfa error	Beta error	CRT		HRT	α/β= 1.5	α/β= 3.0	α/β= 1.5		α/β= 3.0	α/β= 1.5
Pollack et al. [5,6]	2006, 2013	Y	Y	0.05	90%	38 × 2 Gy		26 × 2.7 Gy	177 Gy	127 Gy	197 Gy		133 Gy	84.2 Gy	^a
Marzi [7]	2009	Y	Y	0.05	80%	NA		NA	NA	NA	NA		NA	NA
Hoffman et al. [8]	2013	Y	Y	NA	NA	42 × 1.8 Gy		30 × 2.4 Gy	166 Gy	121 Gy	187 Gy		130 Gy	80.2 Gy
Norkus et al. [9]	2013	N	Y	0.05	80%	38 × 2 Gy		20 × 3.15 Gy	177 Gy	127 Gy	195 Gy		129 Gy	78.5 Gy
Dearnaley et al. [12,13]	2012, 2016	Y	Y	0.05	80%	37 × 2 Gy		20 × 3 Gy	173 Gy	123 Gy	180 Gy		120 Gy	77.1 Gy	^a
Aluwini et al. [15,16]	2015, 2016	Y	Y	0.05	92%	39 × 2 Gy		19 × 3.4 Gy	182 Gy	130 Gy	211 Gy		138 Gy	90.4 Gy	^a
Lee et al. [17]	2016	Y	Y	NA	NA	41 × 1.8 Gy		28 × 2.5 Gy	162 Gy	118 Gy	187 Gy		128 Gy	80.2 Gy
Catton et al. [18]	2017	y	y	0.05	85%	39 × 2 Gy		20 × 3 Gy	182 Gy	130 Gy	180 Gy		120 Gy	77.1 Gy
Arcangeli [10,12]	2010, 2017	Y	Y	NA	NA	40 × 2 Gy		20 × 3.1 Gy	187 Gy	133 Gy	190 Gy		126 Gy	81.5 Gy	^a

^aFor the same analysis, we considered the one with the longer follow-up.

RCT: randomized controlled trial; NA: not available; IS: initial sample; FS: final sample; Y: yes; N: no; GU: genitourinary; GI: gastrointestinal; HRT: hypofractionated radiotherapy; CRT: conventional radiotherapy.

Articles were selected for inclusion in the systematic review if they evaluated data from patients with localized prostate cancer treated with CRT or HRT. Randomized clinical trials with follow up of at least three months were included [5–18]. Revisions and studies that did not report the outcomes of interest or adequate data were excluded [1,19–28].

When duplicate reports of the same study were found, data from the most complete report was included in the analysis. In studies of the same population but with different endpoints, we used the most recent article that offered the endpoint of interest [5,6,10–16].

Data extraction

Data was independently extracted from each report by two authors (FKC and WB) according to the ‘Preferred Reporting Items for Systematic Reviews and Meta-analysis Statement (PRISMA)’ [29], using a data recording form developed for this purpose. The data was reviewed and compared by a third author (AC). Instances of disagreement between the two extractors were resolved by reaching a consensus among the investigators. Whenever necessary, additional information about a specific study was obtained by directly questioning the main investigator.

Checklist and risk of bias

Two authors (FKC and WB) have applied the Scottish Intercollegiate Guidelines Network (SIGN) checklists for comparative studies and controlled trials for all articles included (Supplementary Appendix 1).

Definition of endpoints

The endpoints assessed included GU and GI toxicities as defined by the National Cancer Institute Common Terminology Criteria for Adverse Events [30]. The first three months were considered ‘acute events’, while events later than 12 months were considered ‘late events’. Only events greater than grade 1 were considered [31].

Analysis

For all analyses, the longest follow-up for each trial was used. A pooled estimate of the risk difference in the individual studies was calculated using a random-effect model according to Mantel-Haenszel method and the results were graphically represented using forest plots. All of the results were presented with a 95% confidence interval (95% CI). The diamond at the bottom of the plot summarizes the best-estimate results of the meta-analysis, whose width represents the corresponding 95% CI.

The assumption of homogeneity was checked by a Chi-square test with the degrees of freedom (df) equal to the number of analyzed studies minus one. Additionally, the heterogeneity was measured by the I², which describes the percentage of total variation across the studies that is due to heterogeneity rather than chance [32,33]. I² was calculated from the basic results obtained from a typical meta-analysis as I² = 100% × (Q − df)/Q, where Q is Cochran’s heterogeneity statistic. A value of 0% indicates no observed heterogeneity and larger values indicate increasing heterogeneity. If heterogeneity was detected (I² > 50%), a possible explanation was sought. The presence of possible publication bias was investigated by a funnel plot and if a sub-analysis was possible it was performed to control this bias.

Evaluation of possible bias, as described by Daniel et al. [34], was performed by two authors (FKC and WB). All analyses were conducted with Review Manager v5.0 (the Cochrane Collaboration’s Information Management System).

Results

Study selection

Our initial search identified 1376 articles and after the exclusion of 19 duplicate articles, 1357 studies were selected for detailed analysis of abstracts. 1298 studies were excluded since they did not fulfill the inclusion criteria. A detailed analysis of the remaining 61 articles was then performed. Forty were subsequently excluded due to inclusion criteria and three for incomplete data on the study population. Five articles were grouped as one because they used the same population (Figure 1).

Figure 1. Study selection.

Study characteristics

Thirteen randomized trials were included for systematic review; nine studies containing 7317 patients ultimately remained in the analysis [5–18].

Four studies were excluded because they did not supply adequate data [19–22]. Lukka et al. [19] and Yeoh et al. [20] used the ‘National Cancer Institute of Canada’ and ‘Late Effect in Normal Tissue - Subjective Objective Management Analytic (LENT-SOMA)’ toxicities scale, respectively. These are not comparable with other included trials which used Radiation Therapy Oncology Group (RTOG) scale. While some of the nine selected articles contain the same study population, all were kept in the analysis in order to use the various information that each contributed. However, patient populations were not duplicated in any of the analyses (Table 1). The detailed analysis of bias from each study is present in Supplementary Appendix 1.

Synthesis of results

Acute and late GU toxicity

Three thousand four fifty nine patients from six studies were analyzed for acute GU toxicity. At three months after radiotherapy, acute GU toxicity was similar in patients treated with HRT and CRT (32.6 vs. 31.9%; RD 0.00, 95% CI; −0.03,0.03; p = .81; I² = 0%; Figure 2(A)).

Figure 2. Forest plot (A) Acute GU toxicity. (B) Late GU toxicity. (C) Funnel plot- late GU toxicity. (D) Forest plot- analysis of sensitivity: late GU toxicity.

Seven studies and 6874 patients could be analyzed for late GU toxicity. At 12 months or more after radiotherapy, GU toxicity was also similar (28.7 vs. 28.0%; RD −0.01, 95% CI; −0.04,0.03; p = .67; I² = 52%; Figure 2(B)). Due to high heterogeneity, a sensitivity analysis was performed excluding one article [17] and the six articles showed the same results (33.2 vs. 32.5%; RD −0.01; 95% CI; −0.01,0.02; p = .44; I² = 0; Figure 2(C,D)).

Acute and late GI toxicity

Using 3458 patients from six studies, GI toxicity at three months after radiotherapy was higher in patients treated with HRT versus CRT (27.5 vs. 21.9%; RD 0.06, 95% CI; 0.02,0.10; p = .004; I² = 39%; Figure 3).

Figure 3. Forest plot- acute GI toxicity.

Nevertheless, eight studies with 6988 patients showed statistically similar GI toxicity at 12 months or more after radiotherapy in patients treated with HRT and CRT (12.9 vs. 16.2%; RD −0.01; 95% CI; −0.04,0.02; p = .41; I² = 58%; Figure 4(A)). Because of high heterogeneity, a sub analysis was performed. After excluding two articles [13,14], late GI toxicity remained similar between the schemes, HRT and CRT (11.9 = vs. 12.9%; RD 0.00; 95% CI; −0.02,0.02; p = .96; I² = 27%; Figure 4(B,C)).

Figure 4. (A) Forest Plot- late GI toxicity. (B) Funnel plot- late GI toxicity. (C)– Forest plot- analysis of sensitivity: late GI toxicity.

Discussion

Emerging data from randomized trials evaluating post-radiotherapy toxicity in the treatment of localized prostate cancer showed mixed results when comparing HRT with standard schemes. The study reported here provides evidence that HRT has similar acute and late GU toxicity and late GI toxicity compared to conventional therapy, despite a higher rate of acute GI toxicity was seen. Only toxicity and not tumor control outcomes, was selectively analyzed because of the relatively short follow up interval available by which to definitively judge the latter. It is important though to make expeditious use of data which may reasonably drive clinical practice.

It has been hypothesized that prostate cancer has a low alpha/beta ratio which may be positively affected by higher dose per fraction with regard to local control without additional toxicity [4]. Brenner and Hall [35] analyzed two mature data sets, one using external-beam radiation therapy (EBRT) and other permanent radioisotopic implant, to determine the sensitivity of prostatic cancer to changes in fractionation using the linear quadratic model. The estimated alpha/beta ratio was 1.5 Gy for prostate tumor control. In a second publication, Brenner et al. [4] estimated a similar alpha/beta ratio of 1.2 Gy and they concluded that appropriately designed hypofractionation schemes would be expected to maintain acceptable levels of tumor control and late sequelae, but with reduced acute morbidity, besides the logistic and financial advantage of fewer fractions. More recently, Vogelius et al. [36] published a meta-analysis of 13 randomized trials with the aim of establishing the dose response and fractionation sensitivity of prostate cancer treated with different external beam radiation therapy fractionation schemes. Not taking the overall treatment time effect into account, they found an alpha/beta ratio of 1.2 Gy, as low as found by Brenner et al. [4]. When the overall treatment time was taken into account, the alpha/beta ratio was a little higher (2.7 Gy).

Yeoh et al. [20], in one of the first trials that compared HRT with CRT, showed similar results in late GU and GI toxicity at five years post treatment. In another pioneering trial, Lukka et al. [19] confirmed the findings of similar late GU and GI toxicity in both treatment arms, HRT and CRT. Nevertheless, this study did show worse cancer control outcomes in the HRT group, although equivalent dose levels were low and preceded the dose escalation era.

Kuban et al. [37] reported the preliminary outcome and toxicity of a phase III randomized clinical trial which showed no statistically significant difference in late GU and GI toxicity between HRT and CRT schedules. More recently, Arcangeli et al. [12] reported the final results of a phase III randomized clinical trial which maintained the previously published three-year follow-up data, also with no significant difference in late toxicity between the two groups.

These findings are consistent with one meta-analysis and one systematic review [24,26] which were published in 2013 and 2015, respectively and both did not include the results from the most recently published randomized trials. In addition, the meta-analysis included studies [19,20] which used different toxicity scales such as the ‘(Late Effects of Normal Tissues (LENT)’ instead of the ‘National Cancer Institute Common Terminology Criteria for Adverse Events’, therefore it can increase heterogeneity of results. The same problem was seen with the recent meta-analysis by Cao et al. [28] which reported similar acute GI toxicity with HRT and CRT, dissimilar to our analysis. However, this work contained higher heterogeneity (I² > 50%) which was not controlled by sub-analysis as in our study. A more limited report by Franco et al. [27] corroborated our results but included only three randomized clinical trials [5,6,14], one of which was repeated [5,6]. Therefore, the meta-analysis reported here attempts to combine the most up to date, high level data on moderate hypofractionation schemes.

One factor which must be considered when comparing conventional versus hypofractionated schemes, both for tumor outcome and toxicity, is the BED. Of note is that some comparisons were of isoeffective regimens while others not only hypofractionated but also dose escalated. For example, the trial published by Lee et al. [17] compared a lower equivalent dose in 2 Gy fractions (EQD2), 69.6 Gy, with a higher EQD2, 80 Gy, which of course can impact both tumor control and toxicity rates. Surely, this must be considered when interpreting clinical trial results. BEDs for the comparative study arms included in this meta-analysis are presented in Table 1.

Finally, this meta-analysis does have limitations. First, although all included studies utilized moderately HRT schemes, there was heterogeneity in dose per fraction, BED and treatment technique. Nevertheless, this is unavoidable due to the evolution of technology and the subsequent ability and perceived necessity to dose escalate. Second, the included population was different between studies with regard to initial risk stratification, although this would likely be more relevant for tumor control and survival outcomes which were not the focus of this analysis. Another limitation was our paper included only trials that described toxicity as measured by the Common Terminology Criteria for Adverse Events (CTCAE) (i.e., physician reported). Studies reporting patient reported outcome measures (PROM) were not included and based on this, there is a risk of underestimation of toxicity. The existing studies in the literature did not approach this question and further trials should extend its evaluation by PROMs. Finally, there was some heterogeneity in our meta-analysis, but it was appropriately controlled for through sensitivity analyses.

In conclusion, this systematic review and meta-analysis suggests that acute and late GU toxicity are similar for both CRT and HRT treatment schemes. While HRT was associated with higher acute GI toxicity, late toxicity was similar. Additionally, study results on tumor control have begun to show acceptable results for hypofractionated treatment when contemporary dose levels are used. This in conjunction with the added convenience and lower cost denotes that hypofractionated radiation for prostate cancer will become standard therapy.

Acknowledgments

Ícaro T Carvalho and Willy Baccaglini had the same position as first authors in this paper.

Disclosure statement

No potential conflict of interest was reported by the authors.