APPLES AND ORANGES: COMPARISON OF TREATMENT METHODS FOR PROSTATE CANCER USING BIOCHEMICAL RECURRENCE AS AN ENDPOINT

Authors

  • Peter Grimm,

  • Ignace Billiet,

  • David Bostwick


Sir,

We thank Drs Sooriakumaran, Spahn, and Wicklund for their comment [1] to our recent article [2]. We appreciate the opportunity to respond to their comments. The authors have five major concerns and we will address them separately.

1. EXCLUSION OF MANY SURGICAL STUDIES AS BASED ON GLEASON SCORE AFTER THE SURGERY

The surgical studies that were excluded in this review were not primarily excluded because of the Gleason scoring after surgery, but because the risk group stratification was performed after surgery. Post-surgical risk group stratification precludes meaningful comparison of surgical outcomes with other non-surgical methods. While post-surgery stratified articles may be of interest to a post-surgery patient, the data has little or no utility to the pretreatment patient in the pretreatment assessment of potential outcome.

We agree that are limitations to this type of analysis. However, the Prostate Cancer Results Study Group (PCRSG) analysis in no way rejects surgical studies that predict prognosis based on post-surgery pathology. The PCRSG recognised more importantly, that post-surgical-stratified patients results cannot be reasonably used to advise patients preoperatively on their predicted long-term result and were therefore not appropriate for this analysis. The intent of comparison studies is to allow patients and physicians the ability to predict, preoperatively, from pretreatment parameters (e.g. stage, grade, PSA level, risk group) the prognostic possibilities with various treatments.

2. CONCERN ABOUT UNIFORMITY OF RISK STRATIFICATION

The low-risk classification was uniform and consistent, with no deviation in the classic definition of low-risk disease in all the studies (T1–T2b, Gleason score ≤ 6, PSA level of < 10 ng/mL), However we agree with the authors that the intermediate-risk group definition was more varied amongst articles and thus reflected in the variability of reported results. Further analysis will be required to clarify subgroups. Interestingly, high-risk patients treated with surgery have similar prognosis regardless of definition or treatment period. A study conducted by the Cleveland Clinic [3] showed that regardless of definition and treatment period, the prognosis of radical prostatectomy-treated patients with high-risk disease treated was similar.

3. CONCERN ABOUT DIFFERENT TREATMENT PERIODS POTENTIALLY AFFECTING THE STUDY RESULTS

As all studies were reported within the past 11 years, we think this time period reflected as closely as possible long-term, modern treatment results in all treatment methods. Evidence from numerous randomised control trials has shown a dose effect from radiation on biochemical control and cancer-specific survival, which prompted the panel to recommend a minimum requirement of dose of 72 Gy be given for external-beam radiotherapy techniques. It is worth noting that by today's industry standard, 72 Gy is considered as a low radiation dose. On the other hand, surgical technique, while technically superior in the recent years, has not to date, translated uniformly into improved clinical PSA-based results. The treatment periods reviewed generally involved treatments given within the past 15 years, as only papers published after 2000 were allowed, reflecting modern treatment and techniques. Caseload volume may influence any treatment result. However, a minimum number of patients were required for reporting purposes, eliminating small-volume centre reports. The commentators make a reasonable observation about the lack of studies conforming to standardised reporting. One limitation of this study was recognised, that very few robotic surgical articles to date report in a manner that allow for comparisons. One impact of this paper made be that more authors comply with, and more editors demand for, pretreatment stratification and PSA -based results, which would allow for more detailed comparisons to be performed. This is the first of many PCRSG comprehensive comparisons to come. Hopefully, more data will be available to allow for the subgroup analysis suggested.

4. THE COMMENTATORS MENTION THAT CONFOUNDERS COULD EXPLAIN DIFFERENCES

Confounders certainly do exist certainly but they probably exist in relatively equal proportions for treatment groups, with the exception that younger, healthier patients are more likely to receive surgery. Variations in results are represented by the range, which may be explained by confounders. With more data, confounder stratification may be accomplished but were not considered feasible in this review. The commenters mention an allowance for surgical performance, an interesting concept, which is not easily elucidated from current journal articles. We would be interested on how they would purpose abstracting this information.

The D'Amico article [4] was given by the commentators as an example of a good comparison study. The paper was written in 1998, therefore represented therapies considered by the PCRSG as not representative of current modern brachytherapy technique or radiation doses. Using the commenters' previous arguments about confounders, multiple technical deficiencies in this study would make using this information to compare current modern results inappropriate.

5. PSA ENDPOINT VARIATIONS, AND APPROPRIATENESS OF USING PSA AS AN ENDPOINT

The PCRSG recognised various PSA-based endpoints. Lead time bias certainly can affect interpretation in the short term. The general opinion of the expert panel is that with time, all failures eventually satisfy the various PSA definitions for failure. This is one reason the 5-year limit on studies was imposed, to avoid lead time issues as a possible confounding factor.

If a definition of efficacy is to eradicate all of the disease, the endpoint needs to reflect this. Endpoints other than PSA based (e.g. survival) are subject to variables (patient age, comorbidities, aggressive nature of the disease etc.) that may be completely independent of the treatment given. These variables can mute the apparent difference of treatments and may lead to false conclusions about a treatment's efficacy to eradicate all of the disease.

The PCRSG recognised that various endpoints are used to evaluate outcomes and that future PCRSG analysis of endpoints could include metastasis-free survival, cause-specific survival and overall survival. Biochemical-free progression was selected, as it is was agreed by the panel to be less dependent on other variables and best reflected the efficacy of the treatment in eradicating disease.

Ancillary