Showing posts with label database. Show all posts
Showing posts with label database. Show all posts

Sunday, August 28, 2016

Side effects of salvage radiation – difficult to draw conclusions based on database analyses.


The same researchers (give or take a couple), in analyses of an Italian and US databases, investigated whether it could diminish the side effects of radiation by waiting longer after prostatectomy. The conventional wisdom is that waiting longer allows the newly cut tissues more time to heal completely before the onslaught of radiation.

Unfortunately, the methodology they used for their analyses is problematic. I decided to write about it because it illustrates some of the issues inherent in such database analyses, and raises cautions in drawing conclusions from them.

Hegerty et al. used the US SEER/Medicare database to find 3 groups of patients. All of them had adverse pathology (stage T3 and/or positive margins) after prostatectomy in the period between 1995 and 2007. The three groups were:

1.     RP alone, didn’t have RT (RP only) – 4,509 patients
2.     Adjuvant RT within 9 months of surgery (aRT) -  894 patients
3.     Salvage RT, at least a year after surgery (sRT) -  734 patients

Obviously, this is a large database. The three groups differed somewhat with respect to age, pathology, co-morbidities, history of ED, history of urinary and rectal problems, and demographic characteristics.

·      The RP-only group was most likely to be stage T2, lower Gleason score, older at diagnosis, co-morbidities, no ADT, and a history of GI problems.
·      The aRT group was most likely to be stage T3b, high Gleason score, and have had laparoscopic/robotic surgery.
·      The sRT group was most likely to be younger, have no co-morbidities, diagnosed more years ago, have had open surgery, and ADT.

These differences illustrate the first difficulty in this type of analysis: how do the differences among the patient groups before treatment affect how they will respond to treatment? Some researchers in other studies get around this problem by finding matched pairs of patients selected randomly within certain constraints. In this study, the researchers used two different techniques: “propensity score matching” and a “Cox proportional hazards model.” Propensity score matching attempts to compensate, after the fact, for those factors that seem to be correlated with the treatment decision. After “compensating,” the researchers took note of the Medicare claims for procedures, and recorded diagnoses of side effects as some patients progressed from “RP only” to “aRT” or “sRT.”

This raises another problem with this kind of analysis: the diagnoses and side effects were not necessarily related to the treatment the patient had. For example, a patient who had a colonoscopy a year after his salvage radiation, and perhaps had pre-cancerous polyps removed, is recorded as having a “gastrointestinal diagnosis” and a “gastrointestinal procedure” even though they probably had little relation to his treatment.

The other major shortcoming is that there is also no possibility of investigating how serious the side effects were. For example, urinary incontinence treated with an artificial urinary sphincter carries the same diagnosis as incontinence that only requires one pad a day. Erectile dysfunction managed with a Cialis pill is recorded the same as ED  that only responds to tri-mix injections.

So, for what it’s worth, the authors found the following:

Erectile Dysfunction (ED)

The authors note that rates of ED were high in all three groups, and most men who received post-surgery RT had erectile dysfunction prior to RT. Because of the limitations of the database, we can’t determine the seriousness or causes of the ED complications.

Gastrointestinal (GI)

There were many gastrointestinal procedures. There were more GI events among those who were treated with radiation vs. those who were not. Inexplicably, aRT was associated with lower rates of GI events than sRT. This, again, makes me suspicious that what is observed is an artifact of the methodology rather than a real effect.

Genitourinary (GU)

Men who had radiation were more likely to be diagnosed with incontinence and other urinary side effects than men who had no radiation. But earlier radiation treatment was not associated with worse effects than later treatments. Was this because patients were selected for earlier treatment had early signs of GU recovery after surgery?

Showalter et al. also conducted a similar analysis of an Italian database. They also found that radiation after prostatectomy was associated with worse GI and GU outcomes, but found that outcomes did not improve by waiting longer before radiation treatment.

Because of database limitations, it is extremely difficult to draw meaningful conclusions from these studies. There is no way to judge how serious any of the effects were, or if they were even related to treatment.

Dr. Zelefsky, apparently in agreement with Dr. D’Amico, in a Medscape opinion piece states:
these two studies provide further confidence that earlier treatment should not increase the risk of treatment-related toxicities.”

While I agree that most of the healing of cut tissues from the surgery that will take place has already occurred within a few months, I don’t believe the data in these studies support their conclusions. Until we get better data from randomized clinical trials, how soon to treat with radiation, if at all, remains a difficult decision for both patient and doctor, with no easy answers.