Radiation Therapy for Localized Prostate Cancer: an Update

1. Benefits and harms of radiation therapy for clinically localized prostate cancer compared to no treatment or no initial treatment i.e., watchful waiting in terms of clinical outcomes.
2. What the benefits and harms of different forms of radiation for clinically localized prostate cancer were in terms of clinical outcomes comparing stereotactic body irradiation vs. classically fractionated external beam including 3D vs. IMRT as well as particle therapy and also including high dose rate brachytherapy as well as low dose rate brachytherapy.
3. How do patient's specific characteristics such as age, race or ethnicity, presence or absence of co morbidities affect the outcomes of different forms of radiation?

The results of this analysis revealed essentially no adequate data to answer any of the three questions posed. I agree that the data as available does not answer the questions posed in a prospective randomized way and clearly more data utilizing prospective randomized trials is needed to get answers to these questions, if one only considers low risk prostate cancer patients. Currently the START cooperative group trial between the United States and Canada looking at low risk patients and randomizing them to watchful waiting vs. their choice of a form of treatment surgery vs. radiation options would be an excellent one for nationwide support. But this trial only addresses patients with low risk disease.

The results of this analysis unfortunately did not include patients with T3 disease many of whom have localized disease and there is phase III prospective randomized data (i.e., level 1 data) from Widmark et al (Lancet 2009: 373: 301-308) that demonstrates that the addition of radiation to the prostate improves overall survival and prostate cancer specific survival for patients with locally advanced disease. This data did include some patients with T1 and T2 tumors in addition to T3 tumors and should have been part of the analysis because it shows a clear advantage to the use of radiation in terms of prostate cancer survival and overall survival for these patients.

I do feel that this analysis was clear and organized the scope (except for the data listed in the paragraph above) was appropriate. One questions whether out of 1283 articles just how representative the 62 that were chosen are because they represent less than 5% of the published data on the role of radiation for localized prostate cancer yet the conclusions of no absolute answers regarding the questions posed is reasonable if one looks only at low risk patients.

Finally it is of considerable concern that the Executive Summary (ES-6) references the recent New York Times articles on prostate brachytherapy and suggests a lack of quality assurance with regards to LDR brachytherapy and radiation therapy in general. The questions posed and analysis done for this report did not address quality assurance within radiation therapy delivery and thus, this reference to the New York times articles should be deleted from the report. If the scope of this report had investigated quality assurance within radiation therapy delivery, a multitude of studies would have been found which would address these issues in detail.

Thank you for the opportunity to comment on this report.

1. The START cooperative group trial was discussed in the Future Research section of the report.
2. We had a lengthy discussion on Widmark et al trial (Lancet 2009: 373: 301-308). Even though this trial is phase III prospective randomized data, it didn't meet more than one of our inclusion criteria.
   1. Treatments compared didn't meet our inclusion criteria (looking only at direct comparative studies) to be included under any of our key questions, including radiation therapy to no therapy or no initial therapy.
   2. This trial had all patients treated with hormone therapy.
   3. This trial had included only 21% patients with T1/T2 disease.
   4. The analogy of A+B vs. B → A vs. nothing is an inference and not direct evidence.
3. We searched the data sources with a broader net to obtain all the recent relevant articles and then screened them based on our inclusion criteria and key questions. Establishing objective inclusion and exclusion criteria is part of a "systematic" review process to ensure the studies selected are germane to the particular key questions posed. Asking a different set of key questions (but within the same topic) could very well have resulted in a different set of studies. The studies selected are those studies relevant for our set of key questions, which focused on *comparative* evidence, but are not necessarily representative of all the studies within the field.
4. We have adopted your suggestion regarding the reference to New York Times article and deleted it. Nevertheless, a lack of reporting on adverse event(s) with respect to quality assurance of radiation delivery was observed in the studies reviewed.

I am concerned however that the scope of this review is too narrow. Specifically, I am concerned that the focus on T1-2 prostate cancer (and no consideration of T3 disease) precludes considering strong evidence on the role that radiotherapy plays in the curative treatment of prostate cancer. A seminal paper published by Widmark et al in 2009 (Lancet 2009; 373:301-308) demonstrated that radiotherapy improved survival in men with locally advanced disease. Approximately 20% of these patients had T1-2 disease. Another large Phase III study will be reported at the 2010 ASCO meeting (NCI-Canada Intergroup Study) that shows the same beneficial effect of radiotherapy on survival in men with locally advanced disease. Restricting this review to T1-2 disease limits the number of randomized trials that can be considered and does not tell the entire story. It is clear that radiotherapy changes the natural history of locally advanced prostate cancer (either in the definitive setting or when given postoperatively) and improves survival.

I am concerned that the review uses a trade name (CyberKnife) throughout. Cyberknife is a specific machine that is being used to deliver stereotactic body radiotherapy (SBRT). It is possible to deliver SBRT with most linear accelerators. It would be better to use the generic term SBRT. The report doesn't use linear accelerator or brachytherapy source trade names and they shouldn't use CyberKnife either.

Thank you for your comments. Our report is an update of a previous AHRQ sponsored comparative effectiveness review; that report restricts to studies on patients with T1 or T2 disease. Given that vast majority of patients diagnosed today have clinically localized disease, and not locally advanced disease, our review of patients with T1-T2 disease provides valuable information about radiation therapy for these patients.

The Widmark et al study (Lancet 2009; 373:301-308) did not meet our inclusion criteria. Please see response to reviewer 1.

Your suggestion on the trade name Cyberknife® is noted. This has been replaced with a standard definition -stereotactic body radiotherapy delivered in one or few fractions (SBRT).

In answering Key Question 2 the authors state that "there were no comparisons between EBRT and HDRBT." There are no randomized trials comparing EBRT to HDRBT monotherapy BUT there are two randomized trials comparing EBRT to EBRT plus a BT boost; LDR in one study (Sathya, J Clin Oncol 2005; 23:1192-9) and HDR in another study (Hoskin, Radiother Oncol 2007; 84:114-120). Each study is small and the EBRT dose is unconventional but the BT boost arm is superior in each (without an increase in toxicity).

In the conclusion section of the executive summary the authors state "Available data also suggest that BT is associated with more genitourinary toxicity… compared with EBRT." This statement is inconsistent with a previous statement from page 4 of the executive summary "Two studies did and two studies did not show that LDRBT was associated with significantly more genitourinary toxicity than EBRT." I favor a statement along the lines of "the observed and patient-reported genitor-urinary toxicities are similar between BT and EBRT". This is what I tell my patients.

In the conclusion section of the executive summary the authors state "EBRT administered as a standard fractionation or moderate hypofractionation does not seem to differ with respect to biochemical control and late genitourinary and gastrointestinal toxicities." The authors should specify what they mean by standard fractionation and moderate hypofractionation. I presume that standard fractionation means 1.8-2 Gy per fraction five days per week; but I do not know what the authors mean by moderate hypofractionation. Do they mean >2.5 Gy per fraction? > 3 Gy per fraction?

The study by Sathya et al was included in our report, within the intra-EBRT modality comparison section. In the study by Hoskin et al, less than 80% of participants had T1-T2 prostate cancer, and therefore did not meet our inclusion criteria.

Thank you for your comments and recommendations. The sentence has been modified after taking into consideration your suggestion.

Thank you for your request for clarification regarding the definition of standard fractionation and moderate hypofractionation. We had added the definition of "standard fractionation" in the final report.

Cyberknife is a commercially available radiation therapy machine. In the description of different radiation therapy "types", it is presented as a particular treatment under the heading of "SBRT". This clearly inadequate. Although the Cyberknife is mostly utilized to deliver a few fractions per treatment course, therefore mostly going under the heading of SBRT, the treatment schedules and definitions of IMRT/IGRT/SBRT should not include individual radiation delivery devices. Cyberknife is not a different type of treatment, it's still a radiation-delivery device. For example, Cyberknife should be even listed in Figure 1. I would strongly suggest removing comments such as "stereotactic body radiation (including CyberKnife®)" from the entire document.

The definition of SBRT is completely arbitrary with 5 fractions of less being defined as SBRT. There should be more discussion is clarifying that there is no clear distinction at 5 fractions…

Similarly, Fig 2 suggests that IGRT is a treatment whereby Intensity modulation is also included. Image guidance (IG) has nothing to do with radiation delivery, but is simply the aiming process. Therefore, for ANY radiation delivery technique (2D to IMRT), image guidance could still be used. Including IGRT as a separate "treatment" does not help at all. This should almost be discussed in a different context.

The conclusions are very reasonable. The current state of the data on radiation therapy for prostate cancers does not support any conclusions beyond what the authors have discussed. The authors have done an excellent job at digesting the available information. As such, unfortunately, they are very bland and somewhat useless, if the aim was to determine necessity to treat or differences between treatments.

Only one suggestion:
The authors correctly note in the discussion that the natural history of prostate cancer is so long that conclusions are difficult to make with current trials. For example, the two trials that are most interesting in this context are the Canadian-sponsored START trial planning to enroll 2,130 men, with estimated primary completion date of April 2023 NCT00499174), and the British ProtecT trial that's been running since 1999 planning to enroll 2,050 men, with estimated primary completion date of December 2013 (NCT00632983). The Canadian trial will not have answers till 2023! Even these trials with 2000 patients will be too small to determine significant differences in outcome between treatments, and the techniques currently utilized will be significantly criticized whenever they are published. Much larger number of patient outcomes should be available for analysis for decent decision making; 2000 is not enough.
Particularly in the context of prostate cancer, discussions should include alternative methods of evaluations other than randomized controlled trials, such as mandates of patient outcome reporting and high quality registry studies. Another related discussion should be a study of the factors that affect acceptance of results of trials, even studies considered "high quality".
This might seem outside the context of this particular report, i.e. the discussion of the adequacy of "levels of evidence" the way they are defined today. However, it is actually very relevant in this context. If randomized trials are to be conducted the way they are conducted today and considered as the only acceptable data, the end result will be a dramatic delay of reaching needed conclusions.
The authors also mention at the end the "Future research" section that "Lastly, as has been mentioned earlier, no studies that we reviewed for this technology assessment reported safety data related to the delivery of radiation (e.g., errors in planning software, operator errors, machine malfunctions), it is vital that safety in radiation delivery be actively monitored and diligently recorded for every patient undergoing any form of radiation treatment." These types of "studies" cannot be done in traditional ways (i.e. randomized studies). Mandates to report such incidents should lead to databases of patient outcomes/events/etc. This leads again to a need for high quality registries.
This was somewhat a digression, but a very relevant one.

The START trial was discussed in the Future Research section of the report.

Your insight into alternative study designs is appreciated. We did include study designs other than randomized trials in our review of the literature. In fact, the large majority of included studies were not randomized trials, and registry-based studies were also included.

With respect to safety data, again, we included trial designs other than randomized trials. Your comments on mandatory patient databases are interesting, but not within the scope of this review.