The PTV for all cases included the prostate as defined by our prostate magnetic resonance imaging (MRI) protocol, three-dimensionally coregistered with prostate computed tomography (CT) imaging, matching fiducial to fiducial, plus up to 2 cm of contiguous seminal vesicle and a 2-mm volume expansion in all directions, except posteriorly, where the prostate abutted the rectum. In this region, the margin expansion was reduced to zero, justified by CK system targeting accuracy and reports that prostate cancer does not invade posteriorly in the midline beyond Denonvilliers' fascia. Intermediate-risk patients had a 5-mm dorsolateral prostate-to-PTV expansion to account for their increased risk and potential distance of extracapsular extension near the neurovascular bundle (NVB). Typically, the 2-mm margin expansion used in patients with favorable prognosis split the NVB as defined on T1-weighted gadolinium-enhanced MRI, whereas the 5-mm expansion used for patients with intermediate prognosis fully encompassed it This specific MRI sequence was selected to provide prostate capsular definition, apical definition, and NVB visualization while simultaneously creating a void around the implanted gold fiducial markers that enables the most accurate combination of prostate contouring and MRI-to-CT image coregistration for treatment planning. Although T2-weighted MRI gives detailed intraprostatic anatomic information, such as dominant intraprostatic lesion location and transition zone vs. peripheral zone delineation, the tendency of the gold fiducials to disappear within T2 hypointense signal areas has precluded making it a routine part of our CK SBRT treatment planning image fusion procedure. The urethra was identified by insertion of a Foley catheter, which also provided another reference structure to use in MRI-to-CT image coregistration.
It is our hypothesis that the CK may be used to deliver HDR-like dosimetry to the prostate noninvasively. Supporting this hypothesis, our dosimetry comparison between CK SBRT and simulated HDR showed close similarities between them in coverage of the prostate PTV by the prescribed radiation dose, as indicated by similar V100 characteristics. The CK SBRT also created a similar pattern of dose escalation within the prostate peripheral zone compared with HDR, although the absolute peripheral zone radiation dose distribution was greater in the simulated HDR plans, reflecting the physics inverse square law by which extreme radiation dosage is created in immediate proximity to HDR source dwell positions.
Urethra sparing was clearly more effectively accomplished by means of CK SBRT in this study, with 29 of 30 comparisons favoring the CK SBRT plans, typically by a dose difference on the order of 600 cGy. In all except one case, attempts to match CK SBRT urethra dose sparing with simulated HDR treatment plans resulted in deviation in PTV V100 values for the HDR plans to less than the protocol requirement of 95% PTV coverage. Thus, the CK SBRT plans appeared to better maintain the protocol PTV V100 coverage requirement while also respecting the urethra dose limit when compared directly with case-matched, identically contoured, simulated HDR brachytherapy treatment plans. To our knowledge, this finding was not reported by other investigators and therefore requires additional evaluation by other investigators before definitive conclusions may be drawn.
A higher bladder Dmax was obtained with simulated HDR plans, reflecting the proximity of HDR point source dwell positions relative to the bladder, whereas the higher bladder D10 level seen in the CK SBRT plans likely reflects the effect of streaming CK radiation beams through larger bladder volumes. The clinical significance of these observed bladder dosimetry differences is unknown.
Nearly identical rectal wall Dmax values were obtained with CK SBRT and HDR plans, with a slightly lower median rectal mucosa Dmax value observed in CK SBRT plans. With increasing distance from the point of maximum rectal dose exposure (Dmax), progressively larger differences in rectal wall and mucosa radiation dose sparing in favor of CK SBRT were observed, indicating sharper dose falloff beyond the rectal Dmax point with CK SBRT relative to HDR. Because reported HDR rectal morbidity rates tend to be very low it is unclear whether this more rapid rectal radiation dose falloff with CK SBRT will bring an added clinical benefit, although it suggests that the low rectal injury rates observed with HDR should be equaled or even lower with CK SBRT. In this context, it was reported that a greater incidence of Grade 2 or higher rectal bleeding with HDR brachytherapy was obtained when a larger volume of the rectum received low- to moderate-dose radiation (10–50% of prescribed); this is the rectal exposure range at which we observed the largest differential rectal sparing with CK SBRT relative to HDR Again, it should be emphasized that until our intermodality dosimetry findings are evaluated by other investigators, any conclusions regarding the relative rectal-sparing capability of CK SBRT vs. HDR brachytherapy are preliminary.
It should be noted that both CK and HDR radiation dose sculpting platforms are extremely user programmable, which complicates the direct comparison of these radiation delivery modalities. It is possible that some untested combination of HDR brachytherapy catheter configuration and source dwell position instructions would create a more favorable HDR brachytherapy result, although our relative CK SBRT vs. HDR dosimetry observation trends seemed consistent across the range of prostate volumes and catheter configurations analyzed in our study. Likewise, it also is possible that more effective CK SBRT treatment planning and radiation beam collimator selection could create a more favorable CK SBRT dosimetry result. Despite these caveats, the present study clearly shows that treatment plans that closely approximate those used in HDR brachytherapy for patients with prostate cancer may be constructed and delivered using the CK system.
Further intraprostatic CK SBRT dose escalation
Our early observation of dosimetry trends that favored HDR for the volume of PTV exceeding the prescription dose by 25% or more and CK SBRT for urethra sparing prompted us to run second CK SBRT plan iterations in the first 5 patients. For this exercise, the CK planning computer was instructed to match the urethra Dmax of the corresponding HDR plan while relaxing the CK PTV Dmax limitation and maintaining all other dose limitations to see whether this approach would allow CK SBRT plans to more closely resemble HDR V125 and V150 values. This is exactly what we found; second-iteration CK SBRT V125 and V150 measurements much more closely approached the median simulated HDR V125 and V150 values, more effectively matching HDR PTV dose escalation volumes. Bladder Dmax and all rectal dosimetry parameters changed very little in the second-iteration CK SBRT plans compared with the de novo SBRT plans, although there was a more significant and variable increase in the bladder D10 dosimetry statistic, indicating that more stringent attention to bladder sparing may be required if extreme intraprostatic dose escalation is attempted with CK.
In summary, our second-iteration CK SBRT plans show that intraprostatic dose escalation approaching parity with HDR appears possible, although with attendant partial or complete loss of the superior urethra sparing observed in the initial CK SBRT plans. Because urethral strictures were reported with HDR brachytherapy a reasonable strategy might be to escalate intraprostatic dose as much as possible with CK SBRT while still respecting the urethra dose limits of the CK protocol. This is the approach we continue to use in our ongoing Virtual HDRsm CyberKnife clinical trial, which shows increasing PTV V125 and V150 trend lines over sequential patients.
Although our CT and MRI planning sequences are not specifically designed to delineate the peripheral zone, inspection of the 125% (orange) and 150% (red) isodose lines shown in compared with the pathologic illustration suggests a significant degree of coincidence between CK dose escalation zones and the anatomic peripheral zone of the prostate. A more direct approach would be to perform specific peripheral zone and dominant intraprostatic lesion dosimetry comparisons between the modalities from T2-weighted MRI imaging sequences should such images become available for inspection in future patients.
Radiobiologic relevance of intraprostatic dose escalation
Whether the radiation dose delivery platform is CK SBRT or HDR, the prescription dose of 38 Gy in four fractions is the dose calculated to deliver a biologically lethal blow to the cancer; therefore, it is unclear to what extent further dose escalation beyond this level within the prostate may be necessary or beneficial. However, the exact α/β ratio of prostate cancer upon which hypofractionation schedules are calculated remains uncertain, as discussed in the recent report of Williams. If we use the HDR literature to justify CK treatment, an argument may be made that CK practitioners are well advised to mimic HDR intraprostatic dose distribution as closely as possible to maximize the possibility of reproducing the favorable HDR clinical result. Intraprostatic dose escalation beyond the prescribed dose level, as naturally occurs with any form of brachytherapy, provides backup cancer-cell–killing power in the event that cancer-cell populations with higher α/β ratios exist within heterogeneous populations of prostate cancers and patients with prostate cancer. It also should be noted that Lotan obtained the most reliable tumor ablation in prostate cancers in a nude mouse model with a dose of 45 Gy in three fractions, a more aggressive dosing schedule than described in our study or reported in the HDR literature, again making a case for intraprostatic dose escalation.
Treatment delivery and dosimetry accuracy
Even infinite computerized dose-sculpting capability is clinically meaningless unless the targeting accuracy of the delivery system is sufficient to ensure precise delivery of the treatment plan. Unlike typical image-guided radiotherapy systems, which detect and correct the target volume position only once at the beginning of each treatment, the CK robotic delivery system uses a unique stereoscopic X-ray–based tracking system that updates and corrects robotic linear accelerator position with regard to both translational and rotational target volume movements up to 100 or more times per treatment, resulting in submillimeter targeting accuracy for brain and spine applications
Because the fiducial-based CK tracking procedure for prostate treatment is comparable, the delivery accuracy for prostate cancer theoretically should be identical to that described for brain and spine applications, but with the important caveat that prostate motion is potentially more complex to accurately track and correct for than relatively more fixed brain and spinal targets. Movement caused by bowel peristalsis and bladder filling can cause a rapidly shifting, rotating, and even deforming prostate target volume Until the quantitative effects of these added prostate motion and potential deformation complexities are understood in greater detail, this remains a valid point of criticism for investigators who discuss CK system targeting accuracy in the treatment of prostate cancer.
HDR brachytherapy accuracy is also subject to potential error and distortion because of such factors as variable and potentially significant HDR source transit dose contribution that is neglected by HDR planning computers, prostate volume fluctuation caused by needle trauma, and longitudinal HDR catheter translocation during the course of the patient's hospitalization, with the latter likely representing the largest source of potential HDR brachytherapy targeting error.
Regarding the resemblance of the computer-generated treatment plan to the actual delivered treatment, both CK and HDR brachytherapy have potential dosimetry deviations. Because the sources of dosimetry errors and targeting inaccuracy are different between these modalities, their relative magnitudes are speculative, and as such, it is unknown which modality most consistently delivers its treatment plan more accurately.
Our Virtual HDRsm CyberKnife clinical series is small, with maximum follow-up limited to 12 months; however, some preliminary clinical observations may be reported. Our continuously decreasing median 4-month post-CK PSA value of 0.95 ng/ml suggests a similar response slope to that described by the Stanford group, who reported a median 18-month post-CK PSA value of 0.22 ng/ml. On a larger scale of comparison, our observed median 4-month post-CK PSA decrease of 86% appears comparable to the short-term PSA response magnitude reported with other radiation-based approaches, including standard external beam radiotherapy and 103Pd seed brachytherapy, with much longer term follow-up required to assess durability of the response. Early post-HDR brachytherapy PSA responses, similar to the present data, have not been reported to our knowledge. As our series matures, relative PSA-based disease-free survival rates will be compared.
Acute toxicity of Virtual HDRsm CK monotherapy was self-limited and manageable, primarily consisting of several months of α-blocker–dependent irritative/obstructive uropathy, as well as fatigue, and a less than 100% incidence of typically Grade I proctalgia/rectal urgency that usually resolved by 3–4 weeks after CK treatment. Although our early post-CK toxicity data appear very encouraging, definitive toxicity assessment requires significantly longer follow-up because serious radiation-related complications may not manifest until 1 to 2 years posttreatment. Our series is too small and follow-up is too short to make a meaningful statement about the incidence of post-CK erectile dysfunction, although this domain will be assessed in detail as the fully accrued study matures. For long-term toxicity evaluation, the Virtual HDRsm CK monotherapy protocol includes the long-form Expanded Prostate Cancer Index Composite assessment, which measures urinary, gastrointestinal, sexual, and hormonal-mediated sequelae of therapy
Planning target volume coverage by the prescription dose was similar for CK SBRT and HDR plans, whereas percent of volume of interest receiving 125% of prescribed radiation dose (V125) and V150 values were higher for HDR, reflecting higher doses near HDR source dwell positions. Urethra dose comparisons were lower for CK SBRT in 9 of 10 cases, suggesting that CK SBRT may more effectively limit urethra dose. Bladder maximum point doses were higher with HDR, but bladder dose falloff beyond the maximum dose region was more rapid with HDR. Maximum rectal wall doses were similar, but CK SBRT created sharper rectal dose falloff beyond the maximum dose region. Second CK SBRT plans, constructed by equating urethra radiation dose received by point of maximum exposure of volume of interest to the HDR plan, significantly increased V125 and V150. Clinically, 4-month post–CK SBRT median prostate-specific antigen levels decreased 86% from baseline. Acute toxicity was primarily urologic and returned to baseline by 2 months. Acute rectal morbidity was minimal and transient.
It is possible to construct CK SBRT plans that closely recapitulate HDR dosimetry and deliver the plans noninvasively.