Module 6: Protons for Prostate Cancer: A Review of the Evidence
Prostate cancer is the most common malignancy in men, and external beam radiotherapy remains an important treatment option for this disease. Proton beam therapy (PBT) holds some theoretical advantages over photon treatments like intensity-modulated radiotherapy (IMRT). As explained in previous modules, PBT stops within the body and delivers virtually no radiation to normal tissues past the tumor. However, there is uncertainty over where the proton beam stops; this generally requires slightly larger target volumes with PBT. Protons may also be more sensitive to motion of internal organs.
Dosimetric studies in prostate cancer have suggested that PBT reduces radiation to organs at risk compared with IMRT. Several institutions have also reported single-arm clinical outcomes for PBT showing low rates of toxicity. However, the relevant question for patients, doctors, and payers is whether PBT reduces toxicity compared to the dominant photon modality, IMRT. This question has become controversial as all three groups struggle to identify the best role for proton therapy in treatment of prostate cancer. Therefore, the purpose of this article is to review the clinical evidence comparing PBT with IMRT for prostate cancer.
Claims-based evidence
Introduction
Claims-based studies can capture a nationwide sample of patients and leverage the power of large numbers. This may be an important strength when attempting to detect modest absolute differences between modalities. The weaknesses of these studies include exposure misclassification (incorrectly identifying the type of radiotherapy received by patients) and outcomes misclassification (relying on surrogate claims data to represent clinical toxicities). Furthermore, data such as radiation dose, target volumes, and image guidance are often missing from these studies and could be imbalanced in the comparison groups.
Chronic Condition Warehouse
Yu and colleagues reported an analysis from the Chronic Condition Warehouse, a database capturing Medicare claims for certain chronic conditions. Men were assigned to the IMRT group if they had the IMRT planning code or at least 4 codes for IMRT treatment delivery. Men with any code for PBT delivery were assigned to the PBT group, and patients receiving PBT from 6 U.S. centers were included. Toxicities were based on procedure and diagnosis claims (GU: upper urinary tract dysfunction, urethral stricture, incontinence/obstruction, erectile dysfunction; GI: fistula, rectal repair/resection, stenosis; Other: wound complication, damage to skeletal muscle or connective tissue, RBC transfusion, systemic derangements, nerve lesion, fracture).
The authors used a two-to-one matching procedure to adjust for confounders. Factors for matching included "age, race, residence in a metropolitan county, comorbidity, receipt of ADT, prior influenza vaccination or prior visit to a primary care physician (both as proxies for access to care), and income." Toxicities were analyzed at 6 months after treatment for 1,263 men (421 PBT, 842 IMRT) and at 1 year for 942 men (314 PBT, 628 IMRT).
The rate of GU toxicity at 6 months was lower in the PBT group (6% vs 10%) but there were no significant differences between PBT and IMRT at 1 year (19% vs 18%). For GI toxicity, there were no significant differences between PBT and IMRT at 6 months (3% vs 4%) or at 1 year (10% vs 10%). For "other" toxicity, there were no significant differences between PBT and IMRT at 6 months or 1 year.
The authors conclude PBT was "associated with only a modest and transient reduction in genitourinary toxicity" and that otherwise there "was no difference in toxicity… at 12 months post-treatment."
Surveillance, Epidemiology, and End-Results (SEER)-Medicare analysis 1
Sheets and colleagues reported an analysis of claims data from SEER-Medicare. The authors identified 684 men treated with PBT and 6,666 men treated with IMRT. Toxicities were assessed by diagnosis and procedure codes associated the following categories: gastrointestinal, urinary (incontinence and non-incontinence morbidity), sexual dysfunction, and hip fractions. The analysis was limited to toxicities occurring at least 1 year after the end of radiotherapy.
The authors used propensity score weighting to compare the treatment groups. Baseline characteristics were well-balanced with the exception of factors increasing the risk of rectal bleeding (anticoagulation medication or history of arrhythmia/valvular heart disease) which were more common in the IMRT group.
The results showed no significant differences between IMRT and PBT for urinary toxicities, sexual dysfunction, or hip fractions. Gastrointestinal toxicity was more common in men treated with PBT (gastrointestinal procedure rate per 100 person-years: PBT 21.4, IMRT 17.7; gastrointestinal diagnoses rate per 100 person-years: PBT 17.8, IMRT 12.2).
The authors conclude "proton therapy was associated with more gastrointestinal morbidity than IMRT."
Surveillance, Epidemiology, and End-Results(SEER)-Medicare analysis 2
Kim and colleagues reported an analysis of claims data from SEER-Medicare. Men were assigned to the IMRT group if they received either IMRT alone or 3D conformal plus IMRT. Men were assigned to the PBT group if they received either PBT alone or 3D conformal plus PBT or IMRT plus PBT. Toxicities were based on procedure codes, focused on GI problems (GI bleeding, ulceration, fistula, stricture, or colostomy), and were assessed > 6 months from the end of treatment.
The IMRT group comprised 4,645 men and the PBT group 337 men. Notably, this study analyzed patients diagnosed prior to 2005. PBT was "in its relative infancy," and the results showed declines in GI toxicity with PBT over time. Nonetheless, PBT was associated with a higher rate of GI toxicity than IMRT (4-year rate: PBT 8.5%, IMRT 3.3%).
The authors conclude PBT "had the highest GI toxicity of the radiation modalities, although it was very time dependent, and by 2004–2005, GI toxicity associated with protons had decreased significantly."
Summary
In summary, the claims-based studies have not demonstrated a significant, sustained clinical advantage to PBT. The two SEER-Medicare studies suggested worse GI toxicity with PBT compared to IMRT. The Chronic Condition Warehouse analysis is arguably a more robust comparison because it included a more nationally representative sample of proton patients (6 centers versus likely a single center in the SEER-Medicare studies). This report found a modest, transient reduction in GU toxicity with PBT, but no difference in GI toxicity.
Patient-level evidence
Introduction
Compared to claims-based studies, patient-level studies generally examine fewer total cases and therefore have reduced power to detect a difference between PBT and IMRT. However, patient-level studies have advantages. Such studies include more granular data (treatment details, patient health histories/comorbidities, etc.), which allow better controlling of these confounders. The main toxicity endpoints are also directly measured by physician assessment or patient report, rather than relying on surrogate claims data.
University of Pennsylvania
Fang and colleagues reported the University of Pennsylvania experience. Toxicities were prospectively recorded by radiation oncology nurses for 394 men. These toxicities were classified as acute (within 90 days from start of radiotherapy) or late (greater than 90 days from start of radiotherapy) and graded according to the Common Terminology Criteria for Adverse Events version 3.0. The prostate and proximal seminal vesicles were treated to 79.2 Gy with daily image guidance and an endorectal balloon. IMRT was planned with 7 to 9 coplanar beams of 6 or 15 MV energy. PBT was delivered with a passive scatter technique using two opposed lateral beams.
The study employed a 1-to-1 case-matched design on three variables: D’Amico risk group, age, and prior GI or GU disorders. From the total cohort, 94 matches were made (i.e. 188 total patients, 94 treated with IMRT and 94 treated with PBT). The IMRT group was more likely to have hypertension, have received androgen-deprivation therapy, and to report a low pre-radiotherapy GU toxicity score.
The results showed that most men had acute grade 0–1 GI toxicity (IMRT 86%, PBT 96%). A smaller group experienced acute grade 2 GI events (IMRT 14%, PBT 4%), and there were no grade 3 GI toxicities. These differences were not significantly different between IMRT and PBT. Most men had acute grade 0-1 GU toxicity (IMRT 71%, PBT 79%). A smaller group experienced acute grade 2 GU events (IMRT 29%, PBT 21%), and there were no grade 3 GU toxicities. These differences were not significantly different between IMRT and PBT. Late grade ≥2 GI toxicities were uncommon (1-year rate: IMRT 3%, PBT 10%; 2-year rate: IMRT 10%, PBT 14%), as were late grade ≥2 GU toxicities (1-year rate: IMRT 11%, PBT 12%; 2-year rate: IMRT 12%, PBT 13%). After controlling for confounding factors in a multivariable model, there were no statistically significant differences between IMRT and PBT for either acute or late GI or GU toxicities.
The authors conclude "the risks of acute and late GI/GU toxicities did not differ significantly" between PBT and IMRT.
University of Florida
Hoppe and colleagues reported the University of Florida experience. Patient-reported toxicities were prospectively collected at 6 months, 1 year, and 2 years after treatment using the Expanded Prostate Cancer Index Composite (EPIC) questionnaire. The PBT group consisted of 1243 men who received PBT at the University of Florida. The comparator group was 204 men who received IMRT at 9 institutions as part of the Prostate Cancer Outcomes and Satisfaction with Treatment Quality Assessment (PROST-QA) study. The men treated with PBT received 78 – 82 Gy in 1.8 – 2 Gy fractions to the prostate, with or without seminal vesicle coverage. PBT beam arrangement was varied (posterior oblique fields [45%], anterior oblique fields [42%], mixed fields [7%], or opposed lateral fields [6%]). The prostate was stabilized with rectal saline, and daily image guidance was used. The men treated with IMRT received 75.6 – 79.2 Gy in 1.8 – 2 Gy fractions to the prostate, with or without seminal vesicle coverage. For the IMRT cohort, no details about treatment planning or image guidance were given.
The PBT patients were younger, had smaller prostates, were more likely to be Caucasian, were less likely to receive androgen-deprivation therapy, and had higher minimum and maximum PTV doses. When examining outcomes, adjustments were made for age, androgen-deprivation therapy, and prostate size.
The EPIC tool comprises 26 questions in 4 domains: bowel summary, urinary incontinence, urinary irritative/obstructive, and sexual summary. First, the authors examined changes from baseline in the overall scores for each domain. There were no significant differences between IMRT and PBT for any domain at 6 months, 1 year, or 2 years. Next, the authors tabulated "minimally detectable differences" in the EPIC scores from baseline (defined as "differences from pre-treatment values >50% of the standard deviation at any point in time."). There were no significant differences between PBT and IMRT for rates of minimally detectable differences in any domain at 6 months, 1 year, or 2 years with the exception of bowel summary at 6 months (PBT 25%, IMRT 39%). Finally, the authors examined each of 26 individual questions that make up the EPIC tool. PBT had lower rates of "moderate/big" problems with rectal urgency (2-year rate: PBT 7%, IMRT 15%). PBT also had lower rates of "moderate/big" problems with bowel frequency (2-year rate: PBT 4%, IMRT 10%). There were no differences between PBT and IMRT for any of the other 24 individual questions.
The authors concluded "there were no differences in quality of life summary scores between the IMRT and PBT cohorts during early follow-up (up to 2-years). Response to individual questions suggests possible differences in specific bowel symptoms between the 2 cohorts."
Massachusetts General Hospital
Gray and colleagues reported the Massachusetts General Hospital experience.
The PBT group consisted of 95 men who received PBT at Massachusetts General Hospital. The comparator group was 153 men who received IMRT at 9 institutions as part of the Prostate Cancer Outcomes and Satisfaction with Treatment Quality Assessment (PROST-QA) study. Patient-reported toxicities were prospectively collected for the two cohorts using different but comparable tools (PBT: Prostate Cancer Symptom Indices at baseline, 3 months, 1 year, and 2 years after treatment; IMRT: Expanded Prostate Cancer Index Composite at baseline, 2 months, 6 months, 1 year, and 2 years after treatment). The men treated with PBT received 74 – 82 Gy and those treated with IMRT received 75.6 – 79.2 Gy. No details about treatment planning, beam arrangement, or image guidance were given.
The PBT patients were younger and more likely to be Caucasian. Changes in urinary and bowel quality of life from baseline were calculated for each group on a 100-point scale (with 100 representing perfect normal functioning). These changes were examined immediately post-treatment (3 months for PBT, 2 months for IMRT) and at 1 year and 2 years after therapy.
The results showed that immediately post-treatment, patients receiving IMRT had larger declines in bowel functioning, urinary irritation/obstruction, and urinary incontinence compared to PBT (IMRT: mean change -16.0, -16.5, -7.9, respectively; PBT: mean change -1.7, -4.8, -0.9, respectively). However, at 1 year after treatment there were no significant differences between PBT and IMRT in any domain except for urinary irritation/obstruction; PBT had larger declines compared to IMRT (PBT: mean change -6, IMRT mean change – 0.9). At 2 years after treatment, there were no significant differences between PBT and IMRT in any domain.
The authors conclude patients treated with either modality experienced "similar modest quality of life decrements in the bowel domain and minimal quality of life decrements in the urinary domains at 24 months" after treatment.
Summary
In summary, the patient-level studies have not demonstrated a significant, sustained clinical advantage to PBT. Gray and colleagues found modest transient benefits in bowel and urinary symptoms with PBT immediately post-treatment, but these disappeared by 1 year. Hoppe and colleagues detected no differences in summary toxicity scores at any time point; however, on individual questions there was a modest benefit to PBT for bowel urgency and frequency. Finally, Fang and colleagues reported arguably the most robust comparison with a case-matched study of patients treated at a single institution using the same target volumes and delivery techniques. This investigation found no significant differences between PBT and IMRT for any toxicity domain at any time point.
Comparative clinical trial
An open phase III randomized clinical trial is comparing IMRT and PBT (clinicaltrials.gov identifier NCT01617161). This study is enrolling men with low or intermediate risk prostate cancer at 6 U.S. centers. Treatment consists of IMRT (5 to 9 fields) or PBT (opposed lateral fields with either a passive scattering or pencil beam scanning technique) to 79.2 Gy to the prostate and proximal seminal vesicles with daily image guidance. The main endpoint of the study is patient-reported EPIC bowel scores at 2 years after treatment.