SBRT registries (and the interpretation of registry-based data)

Patient registries are potentially a rich source of information with which to evaluate outcomes. They often include patient characteristics, details of the therapies they received, and outcomes tracked over time. They also provide full population data of all patients treated at participating centers, and can provide very large amounts of data over time.

Like a clinical trial, there are specific and uniform definitions used in capturing patient and treatment data, allowing for comparability on a variety of variables. Registries and clinical trials require internal review board (IRB) approval for ethical standards and must comply with HIPAA laws (patients must consent, and patient names are not entered). In the US, they both have an insurance advantage as well: Medicare, Medicaid, and insurance companies may cover the costs of clinical trials and registries for treatments that they would not ordinarily cover. In some situations, they will only provide coverage if the patient is enrolled in a registry or clinical trial.

Unlike a clinical trial, there are usually no detailed patient inclusion and exclusion criteria, and the treatments may vary from center to center and from patient to patient. Because patients are not excluded from the database, registries are capable of providing very large databases for analysis. There is no randomization, so there is selection bias — patients who received different treatments may have been selected for specific reasons. The quality of the data is only as reliable as the clinician entering it, and it is not necessarily subject to peer review as publication of clinical trial results are. As with other large database analyses, it may be possible to find matched cases for control, but that is not the same as randomization. While clinical trials have a hypothesis to be proved or disproved, a registry provides data for quality improvement and for generating hypotheses.

Registries are difficult and expensive to establish and maintain. The American Board of Radiology attempted to create a national brachytherapy registry, but abandoned those efforts in 2015 when issues in its development and implementation “proved to be more daunting and costly than initially anticipated.” In 2012, the American Society of Radiation Oncologists (ASTRO) announced plans to implement a National Radiation Oncology Registry (NROR) with prostate cancer as its first focus. A pilot was completed in June 2015, and there are plans for expansion.

The Registry for Prostate Cancer Radiosurgery (RPCR) was established in 2010. There are 45 participating sites in the US, and the database included nearly 2,000 men as of 2014. They collect three kinds of data for each patient: screening, treatment, and follow-up.

Screening data include age, performance status, rationale for radiosurgery, initial TNM stage, Gleason score, number of positive biopsy cores, use of hormonal therapy, and several baseline measures, including pre-treatment PSA, IPSS, International Index of Erectile Function (IIEF-5) score, Bowel Health Inventory score, and Visual Analog pain score.

Treatment data include radiation delivery device details, treatment dates, dosimetry (e.g., doses, schedules, targets, margins, including doses to specific organs at risk: rectum, bladder, penile bulb, and testicles), and how image tracking was performed.

Follow-up data include periodic tracking of the baseline data collected at screening, as well as physician-reported toxicity. RPCR encourages sites to record follow-up data every 3 months for the first 2 years following stereotactic body radiation therapy (SBRT) and every 6 to 12 months thereafter, for a minimum of 5 years.

Some interim findings have been published by Freeman et al. So far, they have only reported 2-year data on 1,743 patients. Oncological control was reported as biochemical disease-fee survival:

  • Low-risk patients: 99 percent (n = 111)
  • Favorable intermediate-risk patieents: 97 percent (n = 435)
  • Unfavorable intermediate-risk patients: 85 percent (n = 184)
  • High-risk patients: 87 percent (n = 168)

There was no severe late-term urinary toxicity, and one patient developed severe late-term rectal bleeding. Erectile function was preserved in 80 percent of men under 70 years of age, and 55 percent of men over 70.

The other SBRT registry is called the Radiosurgery Society Search Registry (RSSearch Registry) and includes data from 17 community centers treating prostate cancer patients. There were 437 prostate cancer patients enrolled between 2006 and 2015. The data collected is similar to the RPCR Registry. All patients in their first report were treated using the CyberKnife platform (this registry was originated by Accuray, the manufacturer of CyberKnife), although they allowed other platforms in later enrollments.

Davis et al. recently reported their interim findings. Oncological control was reported as 2-year biochemical disease-fee survival:

  • Low-risk patients: 99.0 percent (n = 189)
  • Intermediate-risk patients: 94.5 percent (n = 215)
  • High-risk patients: 89.8 percent (n = 33)

There was no severe (grade 3) acute urinary or rectal toxicity, and very little grade 2. There was no severe (grade 3) late-term urinary or rectal toxicity. The highest incidence of grade 2 late term symptoms was 8 percent with urinary frequency, They did not collect baseline data on sexual function.

Both of these registries are administered by Advertek. The results of the RSSearch Registry were reported in Cureus, which is their own publication. RPCR results were published in Frontiers in Oncology, which is an independently peer-reviewed journal. It is important to note this because questions about the reliability of the data may arise.

If these data look a little too good to be true … well, let’s dig a little deeper. The biochemical disease-free survival figures only reflect 2 years of follow-up. In that short amount of time, many patients have not yet reached their nadir PSA let alone had time for PSA levels to rise 2 points above that nadir. Most of the low-risk patients and many of the intermediate-risk patients would not have had a rise of 2 points in their PSA even if they’d had no treatment.

The toxicity data are very suspect. Unlike a clinical trial where experienced researchers are carefully evaluating patients on a regular schedule, patient evaluations by community clinicians are haphazard. The clinicians may introduce affirmation bias into their assessments — they have incentive to make their numbers look good. The best way to evaluate toxicity is with patient-reported outcomes on validated, guided-response questionnaires, like EPIC. This was not done in either of these registries.

I think SBRT is actually quite a good therapy (I chose it for myself!), but we have to look to other sources for more reliable data. With longer term follow-up the cancer control data from these registries may become more reliable, and may help us generate better hypotheses about which treatment variants work best and on which patient groups.

Editorial note: This commentary was written for The “New” Prostate Cancer InfoLink by Allen Edel.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: