Does radical prostatectomy really extend life … at least in Denmark?


A newly published paper in the Scandinavian Journal of Urology comes to the potentially controversial conclusion that “the gain in life expectancy” after surgical treatment by radical prostatectomy “is minimal” for Danish men with prostate cancer compared to the life expectancy of Danish men in general. In other words, what the authors appear to be suggesting is that surgery as a treatment for most localized prostate cancers (in Denmark in the 15 years prior to 2011) may well have removed the cancer but may also have been having little impact on the overall life expectancy of the majority of patients so treated.

Now, before this conclusion gets endorsed or dismissed out of hand, we need to look at the details in this paper by Røder et al. and some associated facts, so let’s try and do that with some care.

Røder and his colleagues were able to use the national Danish healthcare databases to look at data on all men diagnosed with prostate cancer and treated by radical prostatectomy in Denmark between August 1995 and December 31, 2011 (i.e., over a 15+ year time frame). This comprised a total of 6,489 men — of whom just 22 men were excluded from the analysis. (Most of those excluded were non-Danish patients for whom there was insufficient follow-up data available.) So the actual study is based on data from 6,467 radical prostatectomy patients. (There are no comparable databases in the USA that would allow us to conduct comparable analyses.)

The authors acknowledge, up front, that we have some sound data showing that:

  • Radical prostatectomy as a treatment for localized prostate cancer can lower risk for prostate cancer-specific mortality compared to a simple observational strategy (e.g., watchful waiting).
  • Early detection and treatment of prostate cancer can lower the relative risk for prostate cancer-specific mortality by something like 25 to 44 percent.

However, what we do not have are any data addressing the effects of treatment for localized prostate cancer on actual length of life — and it was this that the authors wanted to study.

To do this, the authors set out to compare the mortality and the survival of the 6,467 patients in the above-mentioned cohort of patients to the mortality and survival of the an age-matched cohort of Danish males in the general population. And it important in this context to give some definitions:

  • Mortality is the number of deaths within a particular time frame. One can calculate both overall mortality rates and disease-specific mortality rates (e.g., prostate cancer-specific mortality).
  • Survival is time from diagnosis or some other specific event to death. In this particular study it was calculated, for each patient, as the time from radical prostatectomy to the patient’s death (for whatever reason).
  • Relative survival is the ratio of observed survival in a population with a specific disorder treated in a particular way or ways to observed survival in a comparator group (in this case an age-matched group of men in the general Danish population).

So here’s what Røder et al. actually found:

  • Of the total cohort of 6,467 men who underwent radical prostatectomy
    • 1,259 (19.5 percent) had D’Amico low-risk disease.
    • 3,118 (48.2 percent) had D’Amico intermediate-risk disease.
    • 1,818 (28.1 percent) had D’Amico high-risk disease
    • 272 (2.2 percent) could not be classified for risk because of incomplete data.
    • 5,093 (78.7 percent) had a Gleason score of 3 + 4 = 7 or less.
    • 1,198 (18.6 percent) had a Gleason score of 4 + 3 =7 or higher.
    • 176 (2.7 percent) had no Gleason score in their records
    • The average (median) age of the patients was 64 years (range, 36 to 77 years).
    • The average (median) PSA level of the patients was 8.8 ng/ml (range, 0.2 to 201 ng/ml).
    • The average (median) follow-up for all patients was 4 years (range, 0.5 to 16 years).
    • 2,293 patients (35.5 percent) were followed for 5 years or more.
    • Only 411 patients (6.4 percent) were followed for 10 years or more.
    • 328 patients (5.1 percent) died during follow-up.
      • 109 patients had died of prostate cancer.
      • 219 patients had died of other causes.
  • When we look at the survival and mortality data for these men
    • The estimated overall 10-year survival was 85.6 percent
    • The 10-year cumulative incidence of prostate cancer-specific mortality was
      • 5.8 percent overall
      • 0.9 percent for low-risk patients
      • 3.4 percent for intermediate-risk patients
      • 12.8 percent for high-risk patients

Table 2 in the full text of the paper offers a highly detailed breakdown comparing the data from this cohort of patients to data from the age-matched, general, male population of Denmark, including relative survival data, excess mortality rate (EMR) per 1,000 person-years, and standardized mortality ratio (SMR). However, here are the bottom line data:

  • The estimated gain in life expectancy at 10 years for men undergoing radical prostatectomy compared to men in the general population was
    • 0.41 years (about 150 days) overall
    • 0.56 years (about 204 days) for low-risk patients
    • 0.42 years (about 153 days) for intermediate-risk patients
    • 0.31 years (about 113 days) for high-risk patients

So there is a small life-expectancy benefit associated with surgical treatment for prostate cancer in this study, and it is reasonable to describe this as “minimal” in a group of men who probably had a reasonable life expectancy of 15 to 25+ years at the time of their surgery, but whether such a benefit is worth the likely loss of erectile and sexual function and all of the other potential and real complications of a radical prostatectomy could only be determined upfront by a well-informed, individual patient — and his decision might well turn out to be wrong!

The real question that has to be determined is whether the data presented by the authors are “meaningful” in an actionable sense. That question is a lot harder to answer, for a variety of reasons listed below (all of which appear to be clearly understood by the authors):

  • The follow-up time in this study is short (at a median of just 4 years). Would we get the same results if the mortality and survival data for these men were compared to those of the male, age-matched Danish population over a median follow-up of more like 25 years (when the majority of the study population had actually died, either of prostate cancer or of something else)?
  • We know that these data from the Danish population are not really comparable to those for (say) the US population over the same time frame because widespread PSA testing was not happening until later in Denmark than the USA and actual PSA screening for risk of prostate cancer has never been endorsed there. (Note the relatively high level of the median PSA at diagnosis and the proportion of the patients being diagnosed with intermediate- and high-risk disease as compared to low-risk disease.)
  • Relative survival data may be susceptible to diagnostic lead-time bias in the same way as actual survival data are.
  • Selection bias probably also affects these data to some extent.
  • There is a complete lack of data on comorbid conditions of the prostate cancer patients treated by radical prostatectomy in this study.

We now know, from several long-term active surveillance cohorts, but most particularly those of the Sunnybrook group in Canada and the Johns Hopkins group in Baltimore, MD, that there is very little risk indeed of prostate cancer-specific mortality (or even prostate cancer metastasis) at 15 years for low-risk and “favorable” intermediate-risk patients managed on active surveillance. It seems increasingly likely to The “New” Prostate Cancer InfoLink that any serious benefits to life expectancy from early and aggressive treatment for localized prostate cancer are likely to be confined to a subset of men with unfavorable intermediate-risk or high-risk disease and limited signs of disease that has spread outside of the surgical specimen … but proving this one way or another may be difficult to accomplish.

The “New Prostate Cancer InfoLink considers this to be an important paper if for no other reason that it clearly demonstrates — yet, again — the difficulty of being able to know exactly what the benefits of early, aggressive treatment really are for an awful lot of men with localized prostate cancer. The situation is very clearly different for men who get diagnosed with node-positive and metastatic disease and may well be very different for most men with other high-risk forms of prostate cancer, but most men today are getting diagnosed a lot earlier than that, and with disease that is significantly less aggressive too.

Editorial note: The “New” Prostate Cancer InfoLink thanks Dr. Martin Andreas Røder for his prompt cooperation in providing us with a full-text PDF copy of this article so that we could look closely at the details of the study.

13 Responses

  1. Does the Study Indicate Time from Diagnosis to RP (which we may as well refer to as “observation time”)?

    The time frame of RPS in this study (1995-2011) nearly overlays the time frame of active surveillance research in neighboring countries, such as by Dr. Fritz Schröder, related to his work on screening as one of the leaders of the European screening trial. It seems likely that some of the men in the Danish healthcare databases or their doctors would have been influenced by that line of thought, as well as the neighboring Swedish work on watchful waiting versus surgery, and might as a result have decided to defer surgery.

    If so, actual (as contrasted with “measured”) “survival” from diagnosis would be longer than in the tables in the study, provided more than a negligible proportion of men were involved for a significant amount of observation time. It seems likely the researchers would not have been able to observe such observation time in the databases. Was there any indication whether they were aware of this issue or whether they addressed it?

    (Obviously there is some period, often short, for scheduling surgery after diagnosis, but this question relates to an intentional longer deferment of surgery.)

  2. Is Follow-up To Short To Get Much Insight From This Study? (Also, a possible issue with risk-wise patient selection)

    It was helpful that both the study and Sitemaster’s comments clarified the follow-up times in this study, as follow-up time is crucial to evaluating whether an approach works in dealing with prostate cancer, in this case radical prostatectomy (RP). In this study, the key fact to me was that “The follow-up time in this study is short (at a median of just 4 years).” Is that simply too little time to be able to see whether an RP for many men has a beneficial effect on survival, especially when “Only 411 patients (6.4 percent) were followed for 10 years or more.”?

    (Context here, keeping in mind Sitemaster’s observation of important differences between Danish and American patients (especially regarding rates of screening), is that the American Cancer Society’s current statistics for survival are that nearly 100% of all kinds of patients are surviving for 5 years, that 99% are surviving for 10 years, and that 94% are surviving for 15 years (with men unfortunately diagnosed with detectable metastasis, detected with older technology, pulling down these success figures, as only 28% of them were surviving to 5 years, based on evidence mostly prior to the advent of a recent string of important drugs and technology for late-stage patients).)

    Again, we can see that the authors acknowledged the follow-up issue, but what have they left us after allowing for the fact that without any doubt short follow-up dilutes any true and meaningful success achieved with an RP in this study?

    (This issue of short follow-up also was prominent in interpretation of the two main studies involved in the screening controversy that confused the US Preventive Services Task Force, which had no urologist or oncologist on its voting panel. The Task Force concluded that treatment had virtually no benefit (in contrast to the clearly more savvy authors of this Danish study), but the Task Force failed to note the follow-up issue in either the PLCO (Prostate, Lung, Colorectal and Ovarian study) or the ERSPC (European Randomized Study of Screening for Prostate Cancer). A number of critics were quickly spotlighting that critical deficiency, noting that, based on arithmetic, follow-up from time of diagnosis in both studies had to be quite short – far shorter than the published time of follow-up that was calculated from the date of enrollment of each patient in the study. However, it was not until the 2013 update of the ERSPC that we learned the actual figures for the ERSPC: a median follow-up of just 6.4 years from diagnosis in the intent-to-screen group versus just 4.3 from diagnosis in the control (no intent-to-screen) group (from the first page of the Discussion section of “Screening and prostate cancer mortality: results of the European Randomised Study of Screening for Prostate Cancer (ERSPC) at 13 years of follow-up” ( http://www.ncbi.nlm.nih.gov/pubmed/25108889 ).)

    Another key issue in this Danish study is selection of patients who would have a substantial chance of benefiting from an RP. We now know what we did not know at all for many years of this Danish study, at least from 1995 to 2002 at the earliest, when the first encouraging results for active surveillance began to trickle out: that many low-risk men, if not the vast majority, are better managed with well-done active surveillance rather than treatment because they will do just fine without treatment, at least for a number of years if not throughout the rest of life. We now in 2015 would not expect to see much of a survival benefit if any for a majority of low-risk men, especially not with short follow-up. Furthermore, our prostate cancer community of patients, researchers and doctors is now trying to come to grips with whether that lack of benefit is also true for men with the lower-risk form of intermediate disease, and it appears to be true that benefit is, at most, limited, based on what we are seeing on this board. (Right?)

    Moving higher up the risk spectrum, especially to disease that is not confined to the prostate or is likely not to be based on overall circumstances (such as a very high Gleason score), there are numerous studies that suggest a differential survival advantage to some form of well-done radiation rather than surgery (though issues like the very recent node-positive debate indicated this is somewhat controversial). It seems to me that we can say at least that RPs for disease that is not confined are very likely to be less beneficial than for those that are confined, and I personally suspect the difference is probably substantial. In sum, it is likely that a proportion of RP patients in this Danish study got treatment that was unlikely to benefit them, and that too would dilute any benefit to the overall group figure. Put another way, better selection of RP patients should yield an improved benefit result for surgery as a treatment.

    Therefore, it is likely that men at both the lower-risk and higher-risk ends of the disease spectrum — some of whom were included in this study — would not be expected to benefit much from an RP; again, better selection of RP patients would yield a better result. I’m curious whether the authors recognized this limitation in the study.

    At least this study helps us firm up the lower boundary of benefit for an RP: not much, if additional lifespan is likely to be just a few more years, which fortunately is not the case for most newly diagnosed patients. Based on other research, it seems reasonable to expect significant benefit from RPs where lifespan is substantially longer and the patient is in the sweet spot where risk is not too low or too high.

    (Sitemaster, I figure that by now I owe you a whole bottle of aspirin.)

  3. Dear Jim.

    As carefully stated in the text above … the paper measured survival not from diagnosis but from the date of surgery until death.

  4. Thanks.

  5. My only comment is that I truly enjoy posts by my friend Jim Waldenfels! You can rely on his accuracy.

  6. Thanks Chuck. It’s good to hear from you. (I may quote you to my wife!)

  7. Mike, you have been hinting at what I have been saying for a long time and that is if men would take care of their overall health, thereby reducing their chance of dying from “other causes”, then they might actually die from prostate cancer rather than with prostate cancer.

    Stan R.

  8. Dear Stan:

    In the end we will all die of something. It is one of the few, utterly certain truths about life. However, it is my suspicion that very few men diagnosed with truly low-risk prostate cancer are ever likely to die of prostate cancer before they die of “something else” (however long they live). The problem is that it can take a long time to really learn, accept, and understand that.

  9. Of course Mike I was referring to otherwise very healthy patients that will benefit from treatment. I have been pushing active surveillance for many years — since Dr Carroll and I presented classes on educating patients and families on over-treatment.

    Stan

  10. I am highly skeptical that this sample size allows adequate statistical precision for estimating effects of prostate cancer treatment on overall survival. What are the standard errors of estimation? How are they estimated?

    Furthermore, stating effects as “average added extra days of life” or “years of life” does not state the distribution of effects, which may be a better way of understanding the tradeoff. Effects of prostate cancer treatment probably are skewed: for most patients, the treatment has no effect on survival, but for some minority percentage, the treatment extends life by many years.

  11. The question arising in my mind, Mike, is whether surgery reduced PCa specific mortality in men with high risk disease. I realize this was not studied, but wonder if surgery extends OS for these men.

    Another statistical question …. since the sample was taken for procedures between 1995 and 2011, how can they measure 10 year survival for anyone after 2005. Did this create a bias in the results? Is this the same point you are making in your comment on 4 yr median follow-up time?

    PS: Specifically, does surgery extend overall survival (OS) for high-risk men versus no treatment or other forms of treatment, as opposed to OS for the general population?

  12. Rick:

    (1) In Table 3 and Figure 2 of this paper, it is clearly shown that, for men with high-risk disease, there was no significant increase or reduction in overall mortality at 1, 5, and 10 years. However, as you have noted, the authors were not studying prostate cancer-specific mortality and the actual/estimated numbers of prostate cancer-specific deaths at 1, 5, and 10 were probably too small to make any meaningful determinations anyway.

    (2) In almost all studies that provide data about survival times of 10 years and longer, what you are actually seeing are actuarial estimations, not actual data. Remember that if you enroll 6,000-odd men in a study like this over 15 years, only about half the patients will be enrolled for more than 7.5 years anyway (with perfect follow-up on all patients), and so there is always a strong actuarial aspect to the analysis. The fact that the median follow-up time in this study was only 4 years only adds to the question of how accurate any actuarial analysis is going to be.

    (3) The question of whether OS is extended by surgery as compared to any other form of treatment or no treatment has never, as far as I am aware, ever been studied in a prospective manner. It was not studied in this trial. In theory such data may become available from the ongoing ProtecT trial in the UK … if the patients are all followed for long enough, which might need to be 20+ years.

  13. timbarik,

    This is not a sample. As the Sitemaster wrote, it is the entire Danish population of men “diagnosed with prostate cancer and treated by radical prostatectomy in Denmark between August 1995 and December 31, 2011.” They are compared to the actuarial survival of the age-matched entire male population of Denmark. So what the study reported are not estimations, they are actuals for the population.

    You are quite correct that some men will live longer with treatment than others.That’s why the study authors looked within risk categories, which is the best way we have so far of stratifying men for probability of treatment success. The low-risk men had the highest survival advantage compared to population expectations (204 days), while high-risk men had only about half that survival advantage. Interestingly, a US clinical trial, called PIVOT, found that surgery provided no survival advantage over watchful waiting, at least within 12 years of follow up. In that study, there was no survival advantage to surgery in the low-risk group, and a small but non-significant (p = 0.07) advantage among higher risk patients.

    Like almost all characteristics in nature, survival, both of healthy and of prostate cancer-afflicted men, is normally distributed. There will always be outliers who survive more or less than the mean. I’m certain there were such outliers within each risk category in Denmark. However, one cannot conclude from that fact that those longer-lived outliers survived longer because of their treatment. The causes (both natural effects and interventions) for longer survival in a subset can only be determined by a randomized clinical trial, and there are many of those. Hopefully, better treatments than just radical prostatectomy will emerge from those trials.

Leave a Reply

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

WordPress.com Logo

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

Google+ photo

You are commenting using your Google+ 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 )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: