A new article in the February 2015 of the Journal of Urology may actually tell us as much (or more) about the difficulties of relying on data from large, retrospective analyses of single institution data as it does about the actual issue references in the headline above.
Michl and colleagues at the Martini Clinic’s Prostate Cancer Center and the Hamburg-Eppendorf University Hospital in Hamburg, Germany, wanted to look into the question of whether post-surgical treatment with a phosphodiesterase 5 (PDE5) inhibitor like sildenafil (Viagra) or tadalafil (Cialis) has any clinical effect whatsoever on risk for biochemical recurrence post-surgery. The entire text of this paper appears to be freely available on line.
Non-clinical data has historically suggested that the use of PDE5 inhibitors post-surgery may actually lower risk for biochemical recurrence, but no significant data had ever previously explored this hypothesis from a clinical perspective. Michl et al. therefore carried out a retrospective analysis of data from > 4,500 patients treated by a nerve-sparing, radical prostatectomy at the Martini Clinic between January 2000 and December 2010.
Here are their core findings:
- The total number of patients treated by radical prostatectomy was 4,752.
- 1,110/4,752 patients (23.4 percent) were subsequently treated with PDE5 inhibitors.
- 3,642/4,752 patients (76.6%) were not treated with PDE5 inhibitors.
- Average (median) patient follow-up was 60.3 months.
- Estimated rates of 5-year biochemical recurrence-free survival were
- 84.7 percent among men treated with PDE5 inhibitors
- 89.2 percent among men not treated with PDE5 inhibitors
- This difference was statistically significant (p = 0.0006).
- Multivariate regression analysis showed that use of PDE5 inhibitors was an independent risk factor for biochemical recurrence (hazard ratio [HR] = 1.38; p = 0.0035).
The authors conclude that, in contrast to the historical, non-clinical data:
… the use of phosphodiesterase type 5 inhibitors after radical prostatectomy may adversely impact biochemical recurrence. Further studies are needed to validate our results.
However, there are very real questions about whether this result reflects “reality” in most clinical practice. In an article about this research on the Medscape Oncology web site, Dr. Michl is quoted as having said that
We were astonished. We expected opposite results.
And he went on to state that the results of the current study should not change clinical practice unless and until they can be substantiated by other research.
There are good reasons for this caution, because this study, based on data from a single institution, which is largely used by a more affluent section of the German population (like certain centers in the USA), could have inherent selection biases that might be affecting the results. For example:
- The study showed no difference in rates of biochemical recurrence between smokers and non-smokers after treatment for prostate cancer, but such a difference has been clearly established in several other studies.
- Age and body mass index (BMI) are also known risk factors for erectile dysfunction, and may facilitate biochemical recurrence after a radical prostatectomy; however, in this study, neither age nor BMI were independently associated with biochemical recurrence-free survival.
- Only a minority of the patients in the current study cohort were actually smokers (but they were distributed equally between the PDE5 and non-PDE5 groups).
In the Discussion section of their article, Michl et al. write that:
It should be emphasized that since the present study is retrospective and includes an observation opposite from what we initially hypothesized based on experimental animal studies and previous clinical observations, it cannot be ruled out that the present findings are the result of chance or unknown confounding variables.
If there is a learning here for patients it is that all data from these types of retrospective analysis, especially when they come from single institutions, have to be looked at with a great deal of caution.
All that we can really tell from this study by Michl et al. is that what seemed like an interesting hypothesis — that PDE5 inhibitor treatment might reduce risk for biochemical recurrence of prostate cancer post-surgery — was not confirmed by this analysis. Based on this study, there seem to still be three possibilities that might be demonstrable in a large, well controlled, randomized clinical trial:
- That the use of PDE5 inhibitors has a relatively small positive effect in reducing risk for biochemical recurrence post-surgery.
- That the use of PDE5 inhibitors has a relatively small negative effect in increasing risk for biochemical recurrence post-surgery.
- That the use of PDE5 inhibitors has no significant effect at all on risk for biochemical recurrence post-surgery.
Whether such a trial will ever be carried out seems rather doubtful. Perhaps a better question to be explored is whether there is a subset of men who — for whatever reason — are highly likely to have a negative response to PDE5 inhibition that significantly increases their risk for biochemical recurrence, so that the use of PDE5 inhibitors can be selectively avoided in such patients following surgery.
Filed under: Living with Prostate Cancer, Management, Treatment | Tagged: biochemical, inhibitor, PDE5, phosphdiesterase 5, recurrence, risk, surgery |
THE "NEW" PROSTATE CANCER INFOLINK 
This study gets my “Association is Not Causation” Award so far this year (but the year is young yet). The way to prove that PDE5 inhibitors cause recurrence would be to randomly assign men who’d had nerve-sparing surgery to receive the inhibitor or placebo, preferably in a double-blind fashion. To me, the selection bias seems obvious — the men who had to take the PDE5 inhibitors probably had a more aggressive form of the disease that nerve-sparing allowed to escape.
We know that the characteristics they used to match the men with — stage, grade, pre-op PSA, age, surgical margin status, and year of surgery — do not fully capture all the risk factors for recurrence. Notably, volume of cancer was not included — a factor that could impact how much nerve sparing could be done, as well as the risk of micrometastases invisible at the surgical margins. Genomic studies have shown there are more aggressive phenotypes.
Neurovascular bundle sparing is a matter of degree — surgeons cut wider into some more than others. If the surgeon thinks that the tumors are too close to the neurovascular bundle, he will often take thin slices into it until he feels he removed it all. Can slicing into a tumor spread it? Possibly. Sometimes he takes frozen sections and has a pathologist standing by, sometimes he does this on judgment, which is risky because small amounts of cancer may be impossible to see. The full text of the study doesn’t say what was done in these patients. It seems entirely plausible to me that when the tumors had grown into the neurovascular bundle, the surgeon cut wider to get it all. This caused more damage to erectile function and those patients were more likely to take PDE5 inhibitors. In this scenario, both the use of PDE5 inhibitors and the return of the cancer had a common cause — nerve sparing with attempted yet inadequate resection of the cancer that had already grown into it.
Hmmm … Well, as indicated in the text, I would certainly agree that the only way to resolve the effect of PDE5 inhibitors on biochemical recurrence is through a randomized clinical trial (which I still think is highly unlikely to happen, for all sorts of reasons).
With respect to exactly why any individual patient in this cohort decided that he would or wouldn’t take a PDE5 inhibitor, I can think of a lot more reasons than need based on extent of surgery. I would be extremely surprised if that was the only reason, and so all those possible reasons also have possible implications with respect to why any individual patient may have had a biochemical recurrence.
I find it astonishing that only a quarter of the cohort got PDE5 inhibitors. Surely almost all would have ED post-prostatectomy?
Two criteria (at least) are missing from the article:
— Linkage between individual surgeons and recurrence. This follows post-surgery as well: “Here is your free sample.” versus “If you really insist.”
— Patient self-sorting pre-surgery: Sex is extremely important to me versus sex is a non-factor to me. A surgeon who did not at least mouth the intent to follow my wishes during the surgery would not operate on me.