OncoLink Editorial: Comparative effectiveness of IMRT, proton and conformal radiation therapy in the treatment of localized prostate cancer
Neha Vapiwala, MD and Curtiland Deville, Jr., MD
The Abramson Cancer Center of the University of Pennsylvania
Last Modified: February 13, 2012
Presenting Author: Nathan Christopher Sheets, Afiliation: University of North Carolina Hospitals, Chapel Hill, NC From: Genitourinary Cancers Symposium 2012 Reporter: Curtiland Deville, MD
Prostate cancer is the most common non-skin cancer in men, and causes the second highest number of cancer deaths annually.
Definitive treatment options for men with organ-confined disease include surgery and multiple radiation techniques, including prostate brachytherapy and external beam radiation modalities, such as 3D-CRT, IMRT, and PT.
The author reports that comparative effectiveness research is urgently needed in prostate cancer because of the rapid adoption of newer and costlier radiation treatments such as IMRT and PT despite limited demonstrated benefit compared to prior technologies.
In this report, the morbidity and disease control outcomes of CRT vs. IMRT are first compared, and secondly IMRT vs. PT, for primary prostate cancer treatment using an observational, retrospective, population-based database analysis.
Surveillance, Epidemiology, and End Results (SEER)-Medicare linked data from 2000 through 2009 for patients with non-metastatic prostate cancer were utilized.
Propensity score adjustment* was used to balance demographic, disease and institutional characteristics for CRT vs. IMRT, resulting in N=6310 vs. N=6666, respectively. Only N=648 were analyzed in the PT group.
Rates of morbidity (gastrointestinal, urinary, erectile dysfunction, hip fractures) and additional cancer therapy (surrogate for recurrence) were calculated using unspecified CPT codes and not reported as toxicity grades.
IMRT use increased from 0.15% in 2000 to 95.9% in 2008, reflecting rapid adoption and utilization.
Comparing IMRT vs. CRT, in propensity score-adjusted analyses, men who received IMRT were:
less likely to be diagnosed with GI morbidity (13.4 vs. 14.7 per 100 person-years, p_0.001) and hip fractures (0.8 vs. 1.0, p_0.006)
more likely to be diagnosed with erectile dysfunction (5.9 vs. 5.3, p_0.006)
less likely to receive additional cancer therapy (2.5 vs. 3.1, p_0.001).
Comparing PT and IMRT, in a propensity-score matched comparison, PT patients had:
a higher rate of GI morbidity (17.8 vs. 12.2 per 100 person-years, p_.001).
No significant differences in rates of other morbidities or additional therapies between PT and IMRT.
IMRT vs. CRT was associated with less GI morbidity and hip fractures, more erectile dysfunction, and less need for additional cancer therapy.
This large-scale population-based study is the first to suggest a simultaneous reduction in disease recurrence and morbidity in patients treated with IMRT vs. CRT for localized prostate cancer.
Proton therapy did not significantly improve outcomes compared to IMRT, but had increased GI morbidity.
These results provide new and long-needed information to decision-makers regarding the currently available evidence on the comparative effectiveness of different RT techniques.
The implications of this study are limited as currently presented, preventing its application and generalizability
Specific morbidities in each category, beyond the general category titles ("GI, urinary, ED, and hip fractures"), were not elaborated upon, other than an indication that CPT codes were used. However, in the printed abstract it states "diagnoses" in the Results section, implying the use of ICD-9 codes. No examples, specific codes, or diagnoses were mentioned.
No timeline for follow-up or median follow-up was provided, only that they included patients from 2000-2009. Thus, it is unclear if median follow-up was similar in each group.
Although this was presented as a population-based study, all PT patients came from a single institution.
The report is limited by its retrospective, observational, nonrandomized nature, and therefore cannot account for inherent bias in:
Patient selection - The presenter reports that the propensity score adjustment was used to account for this issue and balance "demographic, disease, and institutional" characteristics. This seems possible for the 3DCRT vs. IMRT comparison, but not feasible for the IMRT vs. PT comparison, where it was acknowledged that all 684 patients in the proton group came from a single institution based on the location of available proton therapy centers and the SEER-Medicare database. Thus, it is unclear how they were able to balance demographics and institutional characteristics in the PT and IMRT groups. Pragmatically, the "type" of patient traveling to receive proton therapy at a single institution may be inherently different from those patients receiving IMRT at their local institutions throughout multiple states. This is particularly relevant in that patients who travel for a given treatment modality may also have a greater tendency to seek out follow-up surveillance and pursue procedures for treatment-related morbidity.
Treatment-related factors and techniques - the authors acknowledge that these details are not available in the SEER-Medicare database: such as image guidance, target margins, total dose and dose-escalation, field size, etc. Most concerning is that patients receiving multiple modalities were not analyzed separately, and patients receiving both CRT and PT (potentially high-risk prostate cancer) were included in the "PT" group.
The present retrospective study may, at best, serve to be hypothesis- generating once additional detailed methodology, analysis, and results are presented.
A planned randomized trial of IMRT vs. PT, with enrollment anticipated for later this year, should help to better address potential differences between IMRT vs. PT.
* Propensity scores (PS) estimate the predicted probability (propensity) of use of a given drug or procedure in a particular subject, based on his or her characteristics when the treatment is chosen. In principle, the effect of the treatment can then be measured among patients who have the same predicted propensity of treatment, thus controlling for confounding. Use of PS to reduce bias is especially appealing since, under the assumption that all relevant predictors of treatment have been adequately captured, subjects with the same PS should have the same chance of receiving treatment. Therefore, PS are often conceptualized as mimicking randomized trials, although they do so only with respect to factors that have been adequately measured. Randomization, in contrast, removes bias from both measured and unmeasured factors. Taken from: Stürmer T, Joshi M, Glynn RJ, Avorn J, Rothman KJ, Schneeweiss S. A review of the application of propensity score methods yielded increasing use, advantages in specific settings, but not substantially different estimates compared with conventional multivariable methods. J Clin Epidemiol. 2006 May;59(5):437-47. Epub 2005 Oct 13. Review. PubMed PMID: 16632131; PubMed Central PMCID: PMC1448214.