Radiation Oncology in the 21st Century: Prospective Randomized Trials That Changed Practice… or Didn’t!

In a two-part article published in 2009, we discussed the limitations of conventional radiation therapy, the challenges of studying new technologies in radiation oncology, and summarized the state-of-the science for various malignancies (1, 2). Here, we summarize some of the most important prospective, randomized trials that during the intervening years have attempted to improve the tumor control and/or decrease the adverse effects of radiation therapy. For consistency, we have focused here on the null and alternate hypotheses as articulated by the investigators at the onset of each trial, since the outcome of the investigational treatment should be considered clinically significant only if the null hypothesis was rejected. The readers (and patients) are of course free to make their own judgments about the clinical significance of the results when the null hypothesis was not rejected.

In a two-part article published in 2009, we discussed the limitations of conventional radiation therapy, the challenges of studying new technologies in radiation oncology, and summarized the state-of-the science for various malignancies (1,2). Here, we summarize some of the most important prospective, randomized trials that during the intervening years have attempted to improve the tumor control and/or decrease the adverse effects of radiation therapy. For consistency, we have focused here on the null and alternate hypotheses as articulated by the investigators at the onset of each trial, since the outcome of the investigational treatment should be considered clinically significant only if the null hypothesis was rejected. The readers (and patients) are of course free to make their own judgments about the clinical significance of the results when the null hypothesis was not rejected.
Keywords: randomized trials, radiation oncology, survival, local control, adverse events iNTRODUCTiON An overview of recent randomized trials in radiation oncology provides a useful primer on the current state of the field for students, physicians, and researchers. Summarizing the current survival, tumor control, and toxicity data for various disease sites using the highest level of evidence can guide future efforts to improve treatment outcomes for cancer patients. In this article, we included all prospective randomized trials involving radiation therapy that are required to understand the current scope of radiation oncology practice. Many of these trials have influenced clinical practice by either establishing a new standard of care or helping resolve an important question in the management of patients. The article discusses both publically and privately funded clinical trials from across the globe.
Glioblastoma Stupp et al. (3) randomized 695 patients to receive tumor-treating fields (TTFs) or not after the completion of chemoradiotherapy. TTFs are an antimitotic treatment that disrupt cellular division using alternating electric fields applied to a patient's scalp. The null hypothesis was that adding TTF did not prolong progression-free survival (PFS). The null hypothesis was rejected because, at the first interim analysis, the median PFS was 7.1 months with TTF and 4.0 months without [hazard ratio (HR) = 0.62; p = 0.001]. Median overall survival (OS) was also 4.9 months longer after TTF (HR = 0.64; p = 0.004), but even with TTF, one-half of the patients died within 20.5 months, so there is ample room for further improvement.

Anaplastic Glioma and Oligodendroglioma
Cairncross et al. (4) randomized 291 patients with anaplastic oligodendrogliomas and oligoastrocytomas to receive procarbazine, lomustine, and vincristine (PCV) plus radiotherapy or radiotherapy alone after resection. The null hypothesis was that OS was not prolonged by PCV. The null hypothesis could not be rejected because median OS was similar between the two arms (4.6 vs 4.7 years). Secondary analyses, however, showed that, with or without PCV, the 126 patients with 1p19q codeleted tumors lived much longer than those with non-codeleted tumors, and only among those with codeleted tumors, PCV markedly prolonged the median OS compared to radiotherapy alone (14.7 vs 7.3 years; HR = 0.59; p = 0.03).
Van den Bent et al. (5) conducted a randomized trial of PCV plus radiation vs radiation alone following resection in 368 patients with anaplastic oligodendrogliomas. The null hypothesis was that PCV did not prolong OS by 12 months or longer. After a median follow-up of 140 months, the addition of PCV did prolong median OS by 11.7 months (42.3 vs 30.6 months, HR = 0.75; p = 0.018), but since this was a secondary analysis, the null hypothesis could not be formally rejected. Eighty patients had 1p19q codeleted tumors and similar to Cairncross et al. (4), the OS of these patients tended to be longer, especially if they had received PCV.
In the CATNON trial (6), patients with newly diagnosed 1p/19q non-codeleted anaplastic gliomas were randomized in 2 × 2 factorial design to radiotherapy (59.4 Gy in 33 fractions) alone or with concurrent temozolomide, and with or without adjuvant temozolomide. The null hypothesis was that, compared with radiation alone, adjuvant temozolomide did not prolong OS. The null hypothesis was rejected because an interim analysis after enrolling 745 patients showed longer OS with adjuvant temozolomide (5-year OS 55.9 vs 44.1%; HR = 0.65; p = 0.0014). This trial continues for evaluating the benefit of concurrent temozolomide.

Low-Grade Glioma
Buckner et al. (7) conducted a randomized trial in 251 patients with grade 2 astrocytoma, oligoastrocytoma, or oligodendroglioma. The patients were either younger than 40 years and had a subtotal resection or biopsy, or were older than 40 and had a biopsy or resection of any of the tumor; very few patients had total resection. The patients were randomized to radiation alone or followed by PCV. The null hypothesis was that the addition of PCV did not improve the 5-year OS by 21% (corresponding to HR of 0.46). The null hypothesis could not be rejected because, after a median follow-up of 5.9 years, the 5-year OS after radiation plus PCV was non-significantly 9% better (72 vs 63% for radiation alone; HR = 0.72; p = 0.33). An exploratory analysis, after a median follow-up of 11.9 years, however suggested that patients receiving PCV lived much longer (median OS 13.3 vs 7.8 years; HR = 0.59; p = 0.003). Additional exploratory analyses suggested that those with oligodendriogliomas lived the longest, especially if they had received PCV, as did those whose tumors exhibited the R132H mutation in the isocitrate dehydrogenase 1 gene (IDH1 R132H mutation).

Brain Metastases
Brown et al. (8) conducted a randomized trial to compare stereotactic radiosurgery (SRS) alone vs whole brain radiation therapy (WBRT) plus SRS in 213 patients with 1-3 brain metastases; most had lung cancer and about 50% had a single metastasis. The null hypothesis was that cognitive deterioration at 3 months would not be less after SRS alone. The null hypothesis was rejected because the observed rate of cognitive deterioration at 3 months was 63.5% after SRS alone compared to 91.7% after WBRT plus SRS (p < 0.001). Intracranial failure, however, occurred much sooner after SRS alone (HR = 3.6; p < 0.001). After SRS, one-half of the patients died within 10.4 months; therefore, there remains ample room for improvement.
Another trial (9) randomized 194 patients who had undergone resection of a brain metastasis to SRS vs WBRT. Eligible patients had a surgical cavity measuring less than 5.0 cm in maximal diameter and could have up to three unresected metastases. This study had co-primary endpoints of cognitivedeterioration-free survival and OS. The trial was powered to detect a 20% benefit in cognitive-deterioration-free survival at 6 months after randomization, and a 2-month improvement in median OS with SRS. Cognitive deterioration at 6 months was observed in 52% of the patients after SRS compared to 85% after WBRT (p < 0.00031); therefore, that null hypothesis was rejected. There was, however, no improvement in median OS with SRS (12.2 vs 11.6 months; p = 0.70) and, furthermore, intracranial failure occurred sooner after SRS than WBRT (median 6.4 vs 27.5 months; p < 0.0001). It is sobering that the OS of such surgery-eligible patients has remained essentially unchanged since the landmark 1990 Patchell trial (10).

Head and Neck Cancers
A 94-patient randomized clinical trial (11) compared intensity-modulated radiation therapy (IMRT) vs 3D conformal radiation (3DCRT) therapy for tumors of the oropharynx and hypopharynx. The null hypothesis was that grade 2 or worse xerostomia 12 months later was not better after IMRT. The null hypothesis was rejected because such xerostomia was observed in 38% of the patients in the IMRT arm vs 74% in the 3DCRT arm (p = 0.0027). The trial was not powered to reliably assess small differences in locoregional PFS or OS. Figure 3  in 35 fractions). The null hypothesis was that the 3-year PFS was not superior after SBRT. The null hypothesis could not be rejected because after a median follow-up of 37 months, the 3-year PFS was 42% in each arm. Secondary analyses revealed less toxicity after SBRT, however, than after conventional irradiation. Videtic et al. (13) randomized 94 patients with medically inoperable stage I peripheral NSCLC to either 48 Gy in four fractions or 34 Gy in a single fraction. The null hypothesis was that each regimen would have an unacceptable adverse event rate of 17% (grade 3 or worse). The rate of grade 3 or worse adverse events was only 10.3% in the single fraction arm and 13.3% in the 4-fraction arm. Furthermore, the local control rate after 1 year was 97% in the single-fraction arm and 92.7% in the 4-fraction arm. The combination of low toxicity and high tumor control with greater convenience makes the single-fraction option very attractive for further development.
Two prospective randomized trials of SBRT vs lobectomy were started. The STARS trial had a primary endpoint of OS and the ROSEL trial had a primary endpoint of local and regional control. The trials both closed prematurely due to slow accrual. An exploratory pooled analysis of data from both trials (14) included 58 patients and suggested that 3-year OS after SBRT was superior (95 vs 79% after surgery; p = 0.037).

Locally Advanced NSCLC
A randomized trial (15) in 465 patients with locally advanced NSCLC undergoing concomitant chemoradiotherapy tested whether 74 Gy improved OS compared with 60 Gy; unexpectedly, it found that OS was inferior after 74 Gy (median 20.3 vs 28.7 months for the 60 Gy arm; p = 0.008). Median OS with or without cetuximab was similar (25.0 vs 24.0 months) in this same trial that employed a 2 × 2 factorial design. An analysis of dose to cardiac structures revealed numerous cardiac dose volumes including mean pericardium dose were correlated with worse survival, and mean pericardium dose was also associated with grade 3 or higher pneumonitis (16).
Liao et al. (17) randomized 149 patients to receive intensitymodulated photon therapy or 3D-proton therapy. The primary endpoint was treatment failure (TF) rate, a composite of local failure and/or grade 3, or worse pneumonitis. The null hypothesis was that the TF rate at 12 months was not better after proton therapy. The null hypothesis could not be rejected because the 12-month TF rate was non-significantly worse after proton therapy (21.1 vs 17.4% after photons). The mean cardiac dose was lower in the proton arm (p = 0.002) but that did not translate into improved outcomes.
Antonia et al. randomized 713 patients to receive the anti-PDL1 antibody durvalumab or placebo after concurrent chemoradiation (18). The trial had co-primary endpoints of OS and PFS, and it was powered to detect a HR of 0.67 for PFS and a HR of 0.73 for OS. The median PFS was 16.8 months with durvalumab and 5.6 months with placebo (HR = 0.52; p < 0.001); therefore, that null hypothesis (that PFS after durvalumab was not superior to placebo) was rejected. It is too early to determine if OS is prolonged by durvalumab or not and the trial is continuing.

Limited Disease (LD)
Faivre-Finn et al. (19) randomized 547 patients with LD-SCLC to chemoradiation with 45 Gy in 30 fractions (given twice a day) or 66 Gy in 33 fractions (given once a day). The null hypothesis was that the 2-year OS was not superior after 66 Gy. The null hypothesis could not be rejected because 2-year OS was not significantly different (51% after 66 Gy vs 56% after 45 Gy; p = 0.14). The authors suggested that the standard of care should, therefore, remain 45 Gy in 30 fractions (given twice a day).

Extensive Disease (ED)
Slotman et al. (20) randomized 498 patients who had responded to chemotherapy to receive or not receive thoracic radiotherapy to 30 Gy in 10 fractions. The null hypothesis was that thoracic radiotherapy did not improve the 1-year OS by 10% (to 37 from 27%; HR = 0.76). The null hypothesis could not be rejected because, with a median follow-up of 24 months, the 1-year OS after thoracic radiotherapy was not significantly different (33% with thoracic radiotherapy vs 28% without; HR = 0.84; p = 0.066). In secondary analyses, the median OS was the same in both groups (8 months), but the 2-year OS was significantly longer with thoracic radiotherapy (13 vs 3%; p = 0.004).
Consolidative radiotherapy (to the chest and to any extracranial metastases that did not show a complete response after chemotherapy) was evaluated in a 97-patient randomized trial (21). The null hypothesis was that adding radiotherapy did not reduce the risk of death. The null hypothesis could not be rejected and the trial was terminated early for futility because a planned interim analysis demonstrated numerically but not significantly worse 1-year OS after consolidative radiotherapy (50.8 vs 60.1% without radiotherapy; p = 0.21).

Prophylactic Cranial Irradiation (PCI)
Takahashi et al. (22) conducted a 224-patient randomized trial of PCI vs observation for patients with ED SCLC who had no brain metastases on magnetic resonance imaging (MRI). The null hypothesis was that PCI did not improve OS. The null hypothesis could not be rejected and the trial was closed early for futility when a planned interim analysis revealed non-significantly worse median OS in the PCI arm (11.6 vs 13.7 months in the observation arm; HR = 1.27; p = 0.094).

Breast Cancer
In a prospective randomized trial, 4,004 women (with centrally or medially located breast tumors, or laterally located tumors with axillary involvement) received chest wall irradiation with or without regional nodal irradiation (23). The null hypothesis was that regional nodal irradiation did not improve the 10-year OS by 4% (to 79 from 75%). The null hypothesis could not be rejected because, after a median follow-up of 10.9 years, OS was non-significantly improved by 1.6% in the nodal irradiation group (82.3 vs 80.7%; p = 0.06).
A similar randomized trial was conducted in 1,832 women with node-positive or high-risk node-negative breast cancer (24). The null hypothesis was that regional nodal irradiation did not improve the 5-year OS by 5% (to 85 from 80%). The null hypothesis could not be rejected because, after a median follow-up of 9.5 years, OS was non-significantly better by only 1% in the nodal irradiation group (82.8 vs 81.8%; p = 0.38).
These two trials suggested that with regard to OS regional nodal radiation was not superior to treating without it.

Accelerated Partial Breast Radiotherapy
Intraoperative radiation therapy (IORT) following lumpectomy for early breast cancer can spare many patients the need for "standard" whole breast radiation therapy. A 3,451-patient randomized trial compared IORT vs whole breast radiotherapy (25). The null hypothesis was that IORT would lead to worse local control in the irradiated breast (2.5% more local recurrences within 5 years). This null hypothesis was not rejected; local recurrences were more common after IORT (3.3 vs 1.3%; p = 0.042), and the 95% confidence interval (CI) included the pre-specified non-inferiority margin (26).
A similar trial randomized 1,305 women to IORT vs whole breast radiotherapy (27). The null hypothesis was that IORT would lead to worse local control in the irradiated breast (4.5% more local recurrences within 5 years). After median follow-up of 5.8 years, the null hypothesis was rejected. Although there were significantly more local recurrences after IORT (4.4%, 95% CI 2.7-6.1 vs 0.4%, 95% CI 0.0-1.0%; p < 0.0001), the difference was within the pre-specified equivalence margin.
Another randomized trial compared whole breast radiotherapy to interstitial partial breast brachytherapy in 1,184 patients (28). The null hypothesis was that brachytherapy would lead to worse local control in the irradiated breast (3% more local recurrences within 5 years). The null hypothesis was rejected because local recurrences occurred 0.52% (95% CI −0.72 to 1.75) more after brachytherapy (1.44 vs 0.92%), which was within the prespecified non-inferiority margin.

Hypofractionated Whole Breast Radiotherapy
The START-A trial (29) randomized 2,236 women with completely excised invasive breast cancer (pT1-3a, pN0-1, M0) to conventional (50 Gy in 25 fractions) vs hypofractionated (41.6 Gy in 13 fractions, or 39 Gy in 13 fractions) radiotherapy. The null hypothesis was that, after hypofractionation, the local control would be worse (5% more local recurrences within 5 years). After a median follow-up of 5.1 years, the null hypothesis was rejected for both comparisons. The 41.6 Gy arm, 50 Gy arm, and 39 Gy arm had local recurrences of 3.5, 3.6, and 5.2%. The absolute difference in local control at 5 years between the 50 and 41.6 Gy was 0.2% (95% CI: −1.3 to 2.6%) and the difference between the 50 and 39 Gy was 0.9% (95% CI: −0.8 to 3.7%). The 95% CI for the difference excludes the prespecified non-inferiority margin of 5%.
The START B trial (30) similarly randomized 2,215 women to conventional (50 Gy in 25 fractions) vs hypofractionated (40 Gy in 15 fractions) radiotherapy. The null hypothesis was that, after hypofractionation, the local control would be worse (5% more local recurrences within 5 years). The null hypothesis was rejected since a 5% difference in local recurrences was excluded. The rate of local recurrences at 5 years after 40 Gy in 15 fractions was 2.2% compared to 3.3% after 50 Gy in 25 fractions (absolute difference −0.7%; 95% CI: −1.7 to 0.9%).
Longer term results of these two trials (31) confirmed not only that the tumor control was non-inferior after hypofractionated radiotherapy but also that adverse effects were significantly less common after 40 Gy in 15 fractions or 39 Gy in 13 fractions than after 50 Gy in 25 fractions.
Another 1,234 patient randomized trial compared a conventional (50 Gy in 25 fractions) schedule to a hypofractionated (42.5 Gy in 16 fractions) schedule after breast-conserving surgery (32). The null hypothesis was that hypofractionated radiotherapy would result in 5% more local recurrences. The null hypothesis was rejected after a median follow-up of 12 years. The hypofractionated radiotherapy arm had a local recurrence rate of 6.2 vs 6.7% for conventional fractionation (p < 0.001 in favor of non-inferiority).

esophageal Cancer
Van Hagen et al. (33) randomized 368 patients with resectable esophageal or esophagogastric-junction tumors to undergo surgery alone or neoadjuvant chemoradiation followed by surgery. The null hypothesis was that adding neoadjuvant therapy did not improve OS by 6 months (to 22 from 16 months). The null hypothesis was rejected because the median OS improved by 21.4 months (to 49.4 from 24.0 months after surgery alone; HR 0.657; p = 0.003).

Pancreas Cancer
Hammel et al. (34) randomized 449 patients with locally advanced pancreas cancer to chemotherapy vs chemotherapy plus radiotherapy. The null hypothesis was that adding radiotherapy did not increase OS. The null hypothesis could not be rejected because median OS was not significantly different (15.2 months for chemoradiotherapy vs 16.5 months for chemotherapy; p = 0.83). The same trial, using a 2 × 2 factorial design, also found that adding erlotinib to gemcitabine did not prolong median OS. 2,664 received a diagnosis of localized prostate cancer and 1,643 agreed to undergo randomization to active monitoring, radical prostatectomy, or radiotherapy]. The null hypothesis was that prostate-cancer mortality after 10 years of follow-up was not different between the active monitoring and treatment arms. The null hypothesis could not be rejected because, after a median follow-up of 10 years, there was no significant difference among the three arms in the number of deaths from prostate cancer (8 after active monitoring, 5 after surgery, and 4 after radiotherapy; p = 0.48). Furthermore, the OS was also similar among the three arms (p = 0.87).  10-20, or clinical stage T1b-T2b and Gleason Score 7 and PSA < 15). The null hypothesis was that giving the higher dose did not improve the OS. The null hypothesis could not be rejected because, after a median follow-up of 8.4 years, the 5-year OS was nearly identical in the two arms (88 vs 89%) while the 8-year OS rate was also similar (76% in the 79.2 Gy arm and 75% in the 70.2 Gy arm). The patients receiving 79.2 Gy, however, suffered significantly more late grade 2 or worse gastrointestinal and genitourinary toxicity than those receiving 70.2 Gy.

Locally Advanced or Metastatic Prostate Cancer
Abiraterone acetate is a selective, irreversible inhibitor of CYP17, an enzyme that is critical in the production of androgens in the testes, adrenal glands, and prostate-tumor tissue. Inhibition of CYP17 in combination with androgen deprivation therapy (ADT) results in a more effective androgen depletion than can be induced by surgical castration or by GnRH analogs alone. James et al. (40) randomized 1,917 patients with locally advanced or metastatic prostate cancer to ADT alone or ADT plus abiraterone. Local radiotherapy was mandated for patients with nodenegative, non-metastatic disease (N = 537) and encouraged for node-positive disease. Altogether, 786 patients received pelvic radiotherapy. The null hypothesis was that adding abiraterone did not improve the OS. The null hypothesis was rejected because, after a median follow-up of 40 months, there were far fewer deaths in the abiraterone arm (184 vs 262, HR = 0.63; p < 0.001). Subset analysis of the non-metastatic group is awaited.

Metastatic Prostate Cancer
Parker et al. (41) randomized 921 men with castrate-resistant prostate cancer and bone metastases to radium-223 or placebo. The null hypothesis was that radium-223 did not prolong OS with an HR of 0.76. The null hypothesis was rejected because OS was much longer after radium-223 (14.9 vs 11.3 months; HR = 0.70; p < 0.001).

Bladder Cancer
James et al. (42) randomized 360 patients with muscle-invasive bladder cancer to radiotherapy with or without concurrent chemotherapy (mitomycin C plus 5-fluorouracil). The null hypothesis was that adding chemotherapy did not improve the locoregional DFS by 15% (to 65 from 50%; HR = 0.62). The null hypothesis was rejected because, after a median follow-up of 69.9 months, the 2-year locoregional DFS was significantly increased (to 67 from 54%; HR = 0.68; p = 0.03). Five-year rate of OS was 48% in the chemoradiotherapy arm and 35% in the radiotherapy arm (HR = 0.82; p = 0.16). Grade 3 or 4 adverse events were more common in the chemoradiotherapy group.
In this same trial, employing a 2 × 2 factorial design, 219 patients also agreed to be randomized between whole bladder vs partial bladder boost (43). The null hypothesis was that partial bladder boost was 10% worse regarding the 2-year locoregional control rate. This null hypothesis, however, could not be rejected. The 2-year locoregional control rate was worse by 6.4% after partial bladder boost, and the 95% CI included the pre-specified non-inferiority margin. With regard to grade 3/4 toxicity, there was no observable difference between partial bladder and whole bladder boost.

Gynecologic Cancer
Klopp et al. (44) randomized 289 patients with cervical or endo metrial cancer after definitive surgery to conventional radiotherapy vs IMRT. The primary endpoint was a decline on the bowel domain of the Expanded Prostate Cancer Index Composite (EPIC) scale. The null hypothesis was that acute gastrointestinal toxicity as measured by the decline at 5 weeks would not be less after IMRT. That null hypothesis was rejected because patients treated with IMRT had a significantly lower decline in EPIC bowel scores at 5 weeks when compared to conventional radiation (18.6 vs 23.6, p = 0.048). No mention was made by the authors of the null hypothesis regarding tumor control; that information is awaited to feel confident that the tumor control was not compromised.

Rectal Cancer
Ngan et al. (45) randomized 326 patients with T3 rectal cancer within 12 cm of the anal verge to either preoperative short-course radiotherapy (25 Gy in 5 fractions) or preoperative chemoradiation (50.4 Gy in 28 fractions) with concurrent continuous infusional fluorouracil. The null hypothesis was that chemoradiation resulted in 10% less local recurrences by 3 years (15 vs 5%). After a median follow-up of 5.9 years, the null hypothesis was not rejected. The 3-year local recurrence rate was non-significantly 3.1% greater with the 5-fraction regimen (7.5 vs 4.4%; p = 0.24).   The 5-year OS was non-significantly better with the 5-fraction regimen (74 vs 70%; p = 0.62). Bujko et al. (46) randomized 541 patients with cT4 or fixed cT3 rectal cancers to either preoperative 25 Gy in 5 fractions with 3 cycles of consolidation FOLFOX4 chemotherapy or preoperative 50.4 Gy in 28 fractions with concurrent oxaliplatin and boluses of 5-fluorouracil and leucovorin. The null hypothesis was that the 5-fraction regimen did not increase the R0 resection rate by 10%. The null hypothesis could not be rejected and the 5-fraction regimen could not be deemed superior because the R0 resection rate was non-significantly improved only by 6% with the 5-fraction regimen (77 vs 71%; p = 0.07).

DiSCUSSiON
A few generalizations are possible from the above studies. TTFs prolonged the PFS in glioblastoma. In brain metastases, SRS decreased cognitive deterioration compared with WBRT but increased intracranial failures. For head and neck and gynecological cancers, IMRT decreased toxicity, but it remains unclear if it also increased locoregional failures. For early stage NSCLC, a single-fraction SBRT regimen appears very promising. In locally advanced NSCLC, adding durvalumab to radiochemotherapy prolonged PFS, but radiation dose escalation and proton radiotherapy have yet to prove their value. In LD-SCLC, 45 Gy in 30 fractions (given twice a day) remains the standard of care. In ED-SCLC, consolidative radiation therapy did not prove helpful; nor did PCI when there were no brain metastases on the baseline MRI. In breast cancer, IORT could suffice for some patients after lumpectomy and, for those who do require whole breast radiation therapy due to unfavorable pathological features, a hypofractionated regimen is appropriate. In prostate cancer, active monitoring appears to be a reasonable alternative to immediate treatment for clinically localized disease. For those who choose radiation therapy and for those with unfavorable features, there is no proven survival benefit from escalating the radiation dose beyond 70 Gy, but hypofractionated regimens may help decrease treatment burden. Adding ADT to radiotherapy has already prolonged the lives of those with unfavorable features and the addition of abiraterone to ADT appears promising. Table 1 summarizes the areas in Radiation Oncology where there is room for improvement for increasing the local control and/or the survival and/or decreasing the adverse effects of treatment and, we hope, may offer a roadmap for future clinical trials of devices and drugs for improving outcomes. Incorporation of tissue, blood, and imaging biomarkers into those trials will help identify subsets of patients most likely to benefit (or be harmed) by the investigational drug or device.

AUTHOR CONTRiBUTiONS
KZ and BV designed the article. All authors (KZ, CNC, and BV) were involved in writing the article, and all authors (KZ, CNC, BV) approved the final manuscript.