Introduction
External beam radiation has long been a mainstay in the treatment of localized prostate cancer. The use of PSA as a screening tool has lead to an increase in the number of patients diagnosed with prostate cancer, and a shift toward predominance of lower-risk disease. Using PSA-based outcomes has also been associated with higher rates of treatment failure than previously used outcome measures; this has driven therapeutic innovation. Advances such as CT planning, intensity-modulated radiation therapy (IMRT) and increased availability of proton radiation have facilitated more frequent use of conformal prostate radiation, ushering in an era of dose escalation. High doses can be delivered more safely, as improved conformality decreases the volume of bladder and rectum irradiated. Improved efficacy has also been demonstrated, particularly in patients with intermediate-risk disease.
Proton radiation was the earliest available method of dose escalation for prostate cancer. Brachytherapy and IMRT have since evolved as lower cost alternatives. Nevertheless, proton radiation remains an intriguing therapy for localized prostate cancer because of its unique dose-distribution properties. In this article, we review the current role of proton radiation in the management of localized prostate cancer, contrast this therapy with other radiation-based approaches, and highlight the potential of this promising but expensive treatment modality in an era of cost-conscious medicine.
Proton beam radiation
A proton is a charged particle with physical characteristics that make it desirable for use in radiation therapy. Photons, or X-rays, are discrete bundles of electromagnetic energy that have neither mass nor charge. Photon beams deposit dose continuously as they traverse tissue, such that tissue beyond the primary target receives a measurable amount of 'falloff' dose. Proton beams, on the other hand, deposit a large proportion of delivered dose within a relatively short range close to the end of the path of the particle. This high-dose region is called the 'Bragg peak'. Almost no dose is delivered beyond the Bragg peak, the depth of which is determined by beam energy.1 In clinical practice, proton beams of differing energies are combined to broaden the Bragg peak, such that the proximal and distal borders of the primary target region are encompassed. The resulting region that receives a uniformly full dose is referred to as the 'spreadout' Bragg peak (Figure 1). These favorable dose-distribution characteristics mean that critical structures immediately adjacent to target tissue are spared during proton radiation therapy. This modality has been used to successfully manage optic tumors, central nervous system tumors, base-of-skull tumors, and pediatric malignancies.2
Is dose escalation warranted?
Local recurrence of prostate cancer is common after conventional doses of radiation. In two separate Canadian series, Crook et al.3, 4 found locally persistent disease in 31–34% of men who were rebiopsied after receiving 66 Gy; 14–18% of these patients ultimately developed clinically apparent local failure. Although uncontrolled local and pelvic disease is not the predominant cause of morbidity in patients with recurrent prostate cancer, local persistence of disease has been associated with subsequent development of distant metastasis.5, 6, 7 Whether local persistence is the cause of subsequent metastasis or simply a marker of more aggressive disease is not entirely clear, but there are data that show a temporal association between local failure and distant metastasis. This suggests that, whereas in some patients distant disease and local disease progress concurrently and independently, in others locally persistent disease seeds a late wave of metastasis that might be prevented by more effective local therapy.8 It is patients with the latter disease profile who could benefit from higher doses of radiation.
Dose escalation in prostate cancer improves local control and is associated with lower positive rebiopsy rates. A small randomized trial performed at the Massachusetts General Hospital that involved a proton 'boost' in patients with T3 tumors compared a 67 Gy dose with one of 77 Gy. There was a local control advantage for poorly differentiated tumors in the high-dose arm of the study.9 For men with a palpably negative gland 2 or more years after radiation therapy, there was a lower rate of positive rebiopsy in the high-dose arm (28% in the 77 Gy group versus 45% in the 67 Gy group). Zelefsky and colleagues reported rebiopsy data from a cohort of men treated in a dose escalation series; lower rates of positive biopsy were associated with higher radiation doses (7% at 81 Gy, 48% at 75.6 Gy, 45% at 70.2 Gy and 57% at 64.8 Gy).10 Lower rates of positive rebiopsy, improved local control and presumed association between local eradication of disease and disease-free survival support dose escalation as a reasonable strategy for improving outcomes.
Several randomized trials have shown that dose escalation improves prostate cancer-specific outcomes, particularly in patients with intermediate-risk disease (as defined by PSA level, clinical T-stage and Gleason score). Investigators at the MD Anderson Cancer Center randomized 301 patients with T1–3 prostate cancer to 70 Gy or 78 Gy, prescribed to isocenter. After a median of 8.7 years of follow-up, they reported improved biochemical disease-free survival (bDFS) in the high-dose arm (59% for 70 Gy versus 78% for 78 Gy, P = 0.004). The greatest benefit was evident in men whose initial PSA level had exceeded 10 ng/ml (78% versus 39% 5-year bDFS, P = 0.001).11 In a Dutch trial, 669 patients with localized prostate cancer were randomized to either 68 Gy or 78 Gy, prescribed to isocenter. The trialists reported superior bDFS at 5 years for the high-dose group (64% versus 54%, P = 0.02).12 Subgroup analysis revealed that this benefit was limited to patients with intermediate-risk disease.
During a Canadian study 104 men with intermediate or high-risk disease were randomized to either 75 Gy delivered as external beam radiation therapy plus iridium implant or to external beam radiation therapy (66 Gy) only. The researchers detected fewer biochemical failures in the high-dose arm (29% versus 61%, P = 0.0024).13 A subsequent study at Loma Linda/Massachusetts General Hospital involved randomization of 393 men with low-to-intermediate risk disease to 70.2 Gy equivalents (GyE) or 79.2 GyE delivered as a combination of photons and protons. Five-year bDFS was superior in the high-dose arm (91% versus 79%, P <0.001).14 This benefit was evident at all levels of disease risk. Dearnaley reported the results of a UK Medical Research Council trial of dose escalation from 64 Gy to 74 Gy in 843 men with localized prostate cancer who were receiving neoadjuvant hormonal therapy in addition to external radiation; 5-year bDFS was superior in the high-dose arm (71% versus 60%, P = 0.0007).15 To date, no trial has detected a decrease in occurrence of distant metastasis or an improvement in overall survival. The Radiation Therapy Oncology Group (RTOG) 0126 study is powered to detect a survival benefit of dose escalation. For this open trial, patients are being randomized to 70.2 Gy or 79.2 Gy. The accrual goal is 1,520 patients.
Can high doses be delivered safely?
Dose escalation was made possible by technological advances that allowed a greater degree of conformality during delivery of radiation. CT planning facilitated 3-dimensional conformal radiation, wherein the prostate is directly visualized and shaped beams and multiple beam angles are used to target the prostate while minimizing the volume of bladder and rectum in the irradiated field. Prior to this innovation, treatment was planned in two dimensions using bony landmarks to estimate the location of the prostate. Dearnaley and colleagues conducted a randomized trial that compared 2-dimensional radiation to 3-dimensional radiation in men receiving the same dose (64 Gy). The incidences of acute and late radiation proctitis were reduced in the 3-dimensional planning group.16 A phase 1 dose-escalation trial—RTOG 94-06—found a 79.2 Gy dose to be very well tolerated (2.4% incidence of grade 3 late toxicity).17 Zelefsky and co-workers reported excellent tolerance of doses as high as 81 Gy when IMRT was used. IMRT was associated with a significantly lower rate of late grade 2 rectal toxicity than 3-dimensional conformal radiation to the same dose (2% versus 14%, P = 0.005).18 A phase 1/2 dose-escalation study during which 85 men were treated with 82 Gy at 2 Gy per fraction using protons was recently completed at Massachusetts General Hospital/Loma Linda. At a median of 23 months of follow-up, the rate of grade 2 or greater late genitourinary morbidity was 30%. The rate of grade 2 or greater late gastrointestinal morbidity was 12%.19 This may be the highest dose achievable using current proton beam techniques.
In addition to refining methods of dose delivery to the prostate, 3-dimensional planning has increased our understanding of dose–volume constraints for normal tissues such as the bladder and rectum. Constructing a dose–volume histogram facilitates detailed assessment of the dose delivered to an at-risk organ, and generates metrics that are often more predictive of a late complication than the dose delivered to the prostate or the means via which treatment is delivered. Huang and colleagues performed a secondary analysis of the MD Anderson randomized dose-escalation trial and noted that the percentage rectal volume irradiated correlated significantly with the incidence of rectal complications at all dose levels; the absolute volume of rectum was only a factor at doses greater than 70 Gy.20 Cheung et al.21 recently published the results of a normal tissue complication modeling study to predict bladder toxicity after prostate irradiation. They found that patients in whom the 'hottest' 2.9% of the bladder received 78 Gy or more had a higher risk of late grade 2 or greater genitourinary toxicity than those in whom the hottest 2.9% received less than 78 Gy. Currently, dose-volume histograms for IMRT and proton radiation have similar characteristics in the high-dose region. As dose-volume effects are similar, dose-volume constraints should be met such that the safety of dose escalation can be maximized regardless of the type of radiation used.
Proton beam radiation for prostate cancer
The superior depth-dose characteristics of proton beams are suggestive of a dosimetric advantage of protons over photons if similar beam arrangements are employed. This has proven true for tumors in some parts of the body, but might not be borne out for deep pelvic structures such as the prostate. The precise localization of the Bragg peak makes the dose distribution highly sensitive to uncertainties in the particle range within tissue. There is some degree of dosimetric uncertainty at the distal edge of a spreadout Bragg peak, even when the depth of the clinical target has been unequivocally determined. This problem is compounded in prostate cancer patients because of day-to-day variation in the position of the prostate. Although daily imaging allows the isocenter to be shifted so as to prevent mistargeting, adjusting the depth of the spreadout Bragg peak is a far more complex task that is not amenable to daily refinement. Furthermore, isocenter shifts might generate variation in the length of the radiologic path to the target—even in the absence of motion along the axis of the beam—owing to differing bony anatomy in the field introduced by daily tracking of the prostate. To account for prostate motion along the axis of a proton beam, the spreadout Bragg peak must encompass a larger volume, which some practitioners have referred to as a 'field-specific planning target volume'. Variation in bony anatomy requires an additional 'smearing' margin, which further increases this volume. To safely accommodate these additional margins, opposed laterals are the preferred beam angles as they are perpendicular to the rectum. Along this axis, more-generous proximal and distal coverage of the planning target volume does not increase the volume of rectum receiving high doses. Proton beams also have a significant penumbra at depth, which compromises the capacity for sparing of the anterior rectum. These physical constraints curtail the advantages of proton radiation as currently delivered; however, planning studies indicate that constraints might be minimized as intensity-modulated proton radiation is refined.22 Clinical and dosimetric studies will be needed to determine whether the advantage of having no exit dose is matched or even outweighed by the previously described disadvantages.
At present, IMRT is the most widely used form of external radiation for prostate cancer in the US. This technique utilizes multiple nonopposed beams—more than had been used in conventional 3-dimensional conformal therapy. The fluence of each beam is modulated to create a highly conformal dose distribution capable of bending around critical normal structures such as the anterior rectum. IMRT allows the delivery of higher doses to the prostate without a concomitant increase in rectal and bladder toxicity.18 It is the current benchmark to which proton radiation must be compared.
Trofimov et al. recently reported the results of a comparative dosimetric study of 10 patients whose prostate cancer treatment was planned with both IMRT and 3-dimensionsl protons to a dose of 79.2 Gy (79.2 cobalt Gy equivalent [CGE] with protons).23 Proton radiation was planned using opposed lateral fields whereas photon IMRT used seven nonopposed coplanar beams. IMRT yielded better dose conformality, defined as the ratio of the prescription isodose to the volume of the corresponding target. A greater volume of nontarget tissue received the prescription dose with protons. The V70 (volume of bladder that received more than 70 Gy) was 50% higher with protons than with IMRT. The V70 of the rectum was similar. In the low-dose range, IMRT had less favorable characteristics than protons; for example, the rectal V30 for protons was 16–53% lower. The use of a large number of beams during IMRT leads to delivery of low doses of radiation to a large volume of normal tissue. This is referred to as a 'dose bath' (Figure 2).
Figure 2 | The use of a large number of beams during intensity-modulated radiation therapy means that low doses of radiation are delivered to a large volume of normal tissue.
a | Dose distribution for intensity-modulated radiation therapy. b | Dose distribution for a 3-dimensional conformal proton plan. c | The dose difference between a and b, referred to as a 'dose bath'.
Controversies
Risk of secondary malignancy
Both IMRT and proton radiation have the potential to increase whole body radiation exposure. Because IMRT employs multiple treatment fields, it exposes a greater volume of normal tissue to low dose radiation than conventional 3-dimensional conformal treatment. Longer treatment times might also increase total body irradiation as a result of leakage from the treatment 'head'. On the basis of data from atomic bomb survivors, Hall and Wuu24 estimated that men who receive IMRT for prostate cancer would have a secondary malignancy rate of 1.75% at 10 years, compared with 1% for those who receive conventional 3-dimensional conformal radiation. A lower risk of secondary malignancies would be expected to be associated with proton beam therapy as this form of radiation exposes less normal tissue. Miralbell et al.25 performed a dosimetric study in two pediatric patients (one with a parameningeal rhabdomyosarcoma, the other with a medulloblastoma) to assess the influence of the improved dose distribution of protons on the risk of secondary malignancy. They determined that proton radiation had the potential to reduce the risk in the parameningeal rhabdomyosarcoma patient by a factor of more than 2, and in the medulloblastoma patient by a factor of 8–15, as compared with either IMRT or conventional X-rays. An assumption of this study was that the proton treatment would be delivered using a scanned proton beam, wherein secondary neutron production is limited to those generated within irradiated tissue. Most centers in the US, however, currently use passive scattering, whereby the narrow proton beam is allowed to impinge on a scattering foil such that a broad field of useful clinical size is produced. Field-shaping apertures and range compensators are then employed to tailor the beam to the shape of the clinical target. Scattering foils, field-shaping apertures and range compensators are all sources of secondary neutron production.26 Hall and Wuu suggested that the passive scattering method results in an unacceptably high rate of neutron scatter, leading to a greater incidence of secondary malignancy than IMRT. Subsequently, Paganetti et al. reported that the estimates of neutron dose had been grossly overestimated by Hall and Wuu. Pencil-beam scanning (which generates less neutron scatter) and broad-beam scanning result in a lower integral dose than IMRT and probably in decreased rates of secondary malignancy.27 Other investigators have shown experimentally that the neutron dose delivered during broad-beam scanning is much lower than that suggested by Hall and Wuu.28, 29
Secondary malignancy is a major concern in younger populations, but is less of an issue for older patients. There is typically a protracted latency period for the development of a secondary malignancy (10–20 years), which exceeds the life expectancy of many men treated for prostate cancer. The absolute risk of developing a secondary malignancy after radiation treatment for prostate cancer is quite low.30, 31
Cost-effectiveness
Currently, there are five proton facilities operating in the US, three are being constructed and at least ten are in the planning stages. The business model for many of these facilities is predicated upon treating a high number of prostate cancer patients. The cost of these centers ranges from US$25 million to $150 million, depending on the number of treatment gantries installed. The issue of whether the cost of proton therapy for prostate cancer is justified has only recently been raised.
There are no data that indicate a clear clinical advantage of protons over IMRT. Although dosimetric studies have detected lower integral doses associated with proton treatment, there is no difference in any dosimetric measure that has known clinical implications. Further, there are no data to support the notion that protons allow safer dose escalation than IMRT; the bladder neck and anterior rectum receive similar doses to similar volumes when either technique is used. High doses to these structures are the primary barrier to further dose escalation and the primary cause of late morbidity. It has been hypothesized that, although low doses to large volumes of the bladder and rectum have not yet been correlated with treatment toxicity, there could be quality-of-life metrics that are affected by the low-dose bath generated during IMRT. This is an area of active investigation that might be studied in a multi-institutional setting.
It is likely that technological advances, such as intensity-modulated proton therapy, will allow further dose escalations using protons that will not be possible using X-ray-based IMRT. On the basis of this unproven assumption, Konski and colleagues performed an economic analysis to determine whether proton radiation was cost-effective in comparison to IMRT. A Markov model was used to calculate the incremental cost per quality-adjusted life year (QALY) of proton therapy versus IMRT for a man with intermediate-risk prostate cancer. The model assumed that IMRT would be delivered to a dose of 81 Gy and result in a 5-year bDFS of 83%, whereas use of protons would allow delivery of 91.8 Gy and yield a 93% 5-year bDFS with no increase in toxicity. Costs of $58,610 for proton therapy and $25,846 for IMRT were used. The incremental cost-effectiveness ratio was calculated to be $63,578 per QALY for a 70-year-old man and $55,726 per QALY for a 60-year-old man. As the commonly accepted threshold is $50,000 or less per QALY, proton radiation did not seem to be cost-effective, despite generous assumptions regarding achievable dose and efficacy.32 The high cost of proton therapy could be a major stumbling block going forward, unless technical advances can decrease this substantially.
Future directions
Hypofractionation
Currently, radiation therapy for prostate cancer is delivered as a fractionated regimen of 40 or more treatments over an 8–9-week period. This is a cumbersome protocol, particularly for patients who do not live near a treatment facility. Shorter treatment courses would be welcome for both economic and social reasons. In addition, there is a growing body of evidence that prostate cancer is particularly suitable to a shorter course of treatment for biological reasons. Conventional fractionation schemes that deliver 1.8–2 Gy per fraction are based on the premise that tumors are typically less responsive to fraction size than late-responding tissues. The a/b ratio is a measure of fractionation response. Low ratios are typical of late-responding normal tissues; tumors usually have higher ratios. Compared with a high a/b ratio, a low a/b ratio is associated with a greater capacity for repair between fractions and greater relative sparing at small fraction sizes. This implies a superior therapeutic ratio for small fractions for most tumor types. Typical a/b ratios for tumors are 8 or greater, and generally about 3 for late-responding normal tissue. There are, however, tumors that do not adhere to this general rule. These tumors tend to be slowly proliferating or have a low growth fraction—both characteristics of prostate cancer. The a/b ratio for prostate tumors may be as low as 1.5, making them even more sensitive to fractionation than surrounding normal tissue. As a result, the cancer could be spared if small fraction sizes are used.33, 34 The maximum therapeutic ratio of cancer kill to normal tissue injury would be achieved with a regimen of large doses per fraction, suggesting that hypofractionated regimens have a biological advantage.
Hypofractionated external radiation has been used clinically for many years, particularly in the UK. It has been well tolerated, but its efficacy has been difficult to assess as many of the studies of this modality were performed prior to the PSA era. Preliminary data are now available from two more recent randomized trials that compared conventional and fractionated regimens. An Australian study of 120 men compared 64 Gy in 2.0 Gy fractions with 55 Gy in 2.75 Gy fractions, delivered using 2-dimensional techniques; no differences in rectal bleeding or PSA outcome were observed.35 Canadian investigators who studied 936 men treated with either 66 Gy in 2.0 Gy fractions or 52.5 Gy in 2.63 Gy fractions detected a trend toward a lower rate of biochemical nonevidence of disease (bNED) in the hypofractionated arm. The difference between the two groups was not significant, however. It is worthy of note that the biologically equivalent dose in the hypofractionated arm of this trial was lower than that in the conventional arm if an a/b ratio of 1.5 is assumed.36 The RTOG 04-15 trial is currently accruing. Patients will be randomized to 73.8 Gy in 1.8 Gy fractions or to 70 Gy in 2.5 Gy fractions. This is a noninferiority trial designed to detect a 7% difference in bNED.
Despite the limited availability of data, extreme hypofractionation or stereotactic body radiotherapy (SBRT), of only 5–7 fractions at typical doses of 5.5–7 Gy per fraction, is being used in some practices. In the Virginia Mason trial of 40 men, SBRT was used to deliver a dose of 33.5 Gy in 5 fractions of 6.7 Gy each. After a median follow-up period of 48 months, bNED was 90%. Actuarial late genitourinary and gastrointestinal grade 2 toxicities were 16.1% and 9.4%, respectively.37 Extreme hypofractionation might seem a logical extension of more modestly hypofractionated regimens, but it must be studied carefully. Unlike SBRT for lung cancer, the intent of prostate SBRT is not ablative; the urethra is located in the center of the target and must be spared if significant genitourinary morbidity is to be prevented. The effects of larger fraction sizes, in the 5–7 Gy range, might also be poorly predicted by the linear-quadratic model upon which the biological benefit of hypofractionation in prostate cancer is predicated. The potential contributions of reoxygenation and redistribution to tumor control might also diminish when such abbreviated regimens are used.
Despite the need for detailed ongoing investigation, hypofractionation is an exciting development in the management of prostate cancer. The biological benefit needs to be confirmed; the economic and social benefits are clear. Proton radiation could prove to be more advantageous when higher doses per fraction are delivered. Delivery of a more homogenous dose to the target is a characteristic of proton beams, relative to IMRT, in most settings. When the dose per fraction is low, the 'hotspots' associated with IMRT are generally not clinically important. This could change if hypofractionated regimens are more widely used. In the case of SBRT, protons could be the ideal means of delivering high doses per fraction while sparing the urethra. Treatment regimens that use fewer fractions would also minimize the cost of proton beam therapy, increasing its value relative to other modalities.
Partial prostate boosting
Prostate cancer is generally considered to be a multifocal disease, but this is not always the case.38 Even when disease is multifocal, dominant lesions might be more suitable for dose escalation than other areas of the prostate. At present the entire prostate is treated with a uniform dose, as there is no adequate means of 'mapping' cancer within the gland. New imaging techniques, such as dynamically enhanced MRI and 3-dimensional magnetic resonance spectroscopy, could improve anatomic and metabolic evaluation of prostate cancer such that tumor mapping becomes possible.39 The strategy of treating the entire prostate with a conventional dose and delivering a higher dose 'boost' to dominant lesions is already being explored. Several centers have assessed the feasibility of this technique with IMRT;40, 41, 42 intensity-modulated proton radiation would be even better suited to delivering partial prostate boosts.22 This technology is on the horizon. Partial prostate boosting will make issues such as organ motion and uncertainty of proton range even more critical than they currently are.
PET/CT as an in vivo dosimeter
An interesting aspect of proton radiation is the production of positron emitters along the beam path that can be detected using a PET scanner shortly after delivery of a radiation dose. Using PET/CT, in vivo dosimetry can be carried out.43 This unique property of proton beams could be exploited as a quality assurance tool to verify targeting. It might also be used in a form of adaptive therapy; in vivo dosimetry could facilitate adjustment of treatment plans, as proton range uncertainties, penumbra and normal tissue 'spillover' could be studied on a case-by-case basis.
Conclusions
There has been great and sustained interest in proton therapy for pediatric tumors and those of the central nervous system. More recent enthusiasm for this modality relates largely to its potential as a treatment for prostate cancer. Proton radiation has been recognized as an effective tool in the management of prostate cancer, and proffered one of the earliest means of dose escalation. It is now one of several methods used to deliver high-dose conformal radiation to the prostate. There is no hard clinical evidence that—at current levels of sophistication—proton radiation is superior to the more widely used IMRT. Economic analysis indicates that protons, as currently used and priced, are unlikely to be cost-effective; this could change in the future. More economic analyses and quality-of-life studies should be performed before further resources are committed to this promising but costly treatment modality.
Review criteria
The authors drew on their own experience, and personal libraries, to identify information relevant to this Review.
