This is a complicated issue. Bear with me, this is going to be a long post.
For starters, the trial specifics have been published before (Free access):
Stereotactic body radiotherapy (SBRT) is becoming increasingly used in treating localized prostate cancer (PCa), with evidence showing similar toxicity and efficacy profiles when compared with longer courses of definitive radiation. Magnetic ...
www.ncbi.nlm.nih.gov
1. CTV-contouring.
CTV-Contouring was not the same in both arms. Yes, CTV-contouring, it's unbelievable, but sadly true!
From the manuscript:
"A pelvic CT without contrast will be performed for radiotherapy simulation (i.e., treatment planning CT) with a slice thickness of 1.5 mm. For patients enrolled on the MRI-guided SBRT arm, an additional MRI will be obtained in the treatment position on the MRI-guided LINAC."
I have no idea why they did that in the trial. I thought acquiring a planning MRI for prostate RT is standard of care. We do it for 20 x 3 Gy in all patients, I would certainly do it for CBCT-based SBRT, and I have no idea whey they decided not to do it in the CBCT-arm of this trial.
Can anyone think of any reason?
2. Margins.
2mm for MRI-arm, 4mm for CBCT-arm.
From the manuscript:
"
We adopted a smaller margin of 2 mm for the MRI-guided arm given its ability for real-time tracking and superior prostatic anatomy visualization (i.e., greater contour certainty), with little concerns of marginal misses related to excess motion or undercontouring."
...
"Implanted fiducial markers are routinely used to assist with motion management when treating prostate cancer patients with any form of external radiotherapy, including SBRT. These will be considered required for patients treated with CT-guided SBRT except in cases where a medical contraindication is present, as is consistent with our internal SBRT protocol."
...
"Gating-based treatment, with gating based on the imaged location of the prostate organ or on rectal distention, can be employed at the discretion of the treating physician."
They state two reasons for the reduced margins:
a)
Excess motion.
Like many of you noted, excess motion in the CBCT-arm could have been negated by using non-MRI based tools, such as Calypso, Cyberknife, Novalis (or even perhaps the "poor (wo)man's" mid treatment CBCT/kV-image between arcs). Calypso offers real time motion management, Cyberknife and Novalis offer frequent management, the mid-treatment CBCT/kV-image offers the least possibilities. There is no mention in the paper if motion management was allowed in the CBCT-arm, although even the most basic motion management would have been possible with the implanted fiducials. There is also no mention of recommendations on when treatment was stopped in the MRI-arm, in case motion was detected. In theory, one would have to interrupt treatment, once the prostate moved 3mm, since parts of the CTV would then be underdosed. Did they do that? Noone knows.
b)
Undercontouring.
I am not sure if they mean by that the CTV-contouring, since the MRI-guided arm only had initial MRI-assisted CTV-contouring or they mean identifying the prostate contour on the treatment-MRI more precisely than the prostate contour on the CBCT. I do not feel that this is a valid argument in the later case, since with fiducials implanted in the CBCT-arm patients, you do not really need to visualize the prostate, you can simply match to the implanted fiducials. Or am I missing something else?
3. Adaptive planning.
From the manuscript:
"If deemed necessary, online adaptive planning will be performed, wherein the planning GTVs, CTVs, and OARs are deformably transferred to the online MR images via registration, and the treatment plan is reoptimized to meet or exceed the original planning goals."
So, the MRI-guided treatment may lead to better organ sparing and target converage through adaptive planning. This is excellent and one of the main specialties of MRI-treatment, but is also available now on Varian's Ethos, CBCT-based. I am not aware of any comparative studies.
4. Physician and patient biases
The primary endpoint was "
acute grade ≥ 2 GU physician-reported toxicity,, as assessed by the CTCAE version 4.03 scale".
Neither physicians, nor patients were blinded. It's impossible to blind the patients. However, one could indeed have blinded the ones that scored toxicity. One could have utilized blinded study nurses, for instance, that would have scored for acute toxicity.
There is no item such as "acute grade ≥ 2 GU toxicity" in CTCAE 4.03, so I presume they mean a combined endpoint of several CTCAE 4.03 endpoints, including urinary urgency, urinary tract obstruction, urinary frequency. The problem is that many of those automatically are grade 2 as soon as medication is prescribed. And who is prescribing medication (for instance Flomax)? The unblinded physician...
Thus there are 6 potential points that may have impacted the primary endpoint of the trial:
1. CTV-contouring
2. Margins
3. Adaptive planning
4. Motion management
5. Patient bias
6. Physician bias
Saying that treating on the MRI-Linac was the crucial factor, appears to me quite the stretch.
academic.oup.com
