Randomised prostate cancer screening trial: 20 year follow-up
BMJ 2011; 342 doi: https://doi.org/10.1136/bmj.d1539 (Published 31 March 2011) Cite this as: BMJ 2011;342:d1539
All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
We fully endorse Dr. Lewis statement that Prostate Screening can be
bad for health. According to our understanding "Screening" is justifiable
if the prevalence of disease is high, there is lot of suffering of the
patient or mortality is very high and the disease is preventable or at
least amenable to cure. Prostate cancer though prevalent especially in
Europe and United States, more often than not is asymptomatic, slow in
growing and patient eventually dies of some other disease or because of
age. Moreover it is largely chemoresistant and radioresistant. So, Dr.
Lewis had the last word.
Competing interests: No competing interests
We now seem to be straying from the narrow statistical questions of
the original paper on Prostate Cancer screening, into the general
territory of prostate treatment trials. The unassailed fact remains -
Sandblom et al's paper demonstrated an INCREASED prostate-cancer-specific
mortality for those screened.
James Penston perfectly makes my point more generally in
http://www.bmj.com/content/342/bmj.d3369.full
" Beware next wave of propaganda "
I trust he has the last word.
Competing interests: No competing interests
The basic surmise after the results of various studies still leaves a
dilemma that whether treatment or watchful waiting (active surveillance)
after positive results of prostate screening is better than not screening
at all. No screening and/or not knowing the presence of disease would not
cause tension to the patient as "ignorance is bliss" as suggested by Dr.
Lewis.
Several studies (1, 2, 3, 4) now are available to report that
deferred treatment even in patients with localized prostate cancer
postponing or omitting treatment resulted in outcome similar to those
received immediate therapy. Insufficient evidence exists to either support
or refutes the routine use of screening compared to no screening for
reducing prostate cancer mortality. Currently, no robust evidence is
available from randomized controlled trials regarding the impact of
screening on quality of life, harms of screening, or its economic value,
on which this article has put some light.
On the other hand, a randomised trial comparing radical prostatectomy
and watchful waiting management for men with clinically localized prostate
cancer, reported statistically significant reductions in overall
mortality and prostate-specific mortality at 12 years for men receiving
surgery who were younger than 65 years 5. Another study also, in Sweden
concludes that Radical prostatectomy reduces disease-specific mortality,
overall mortality, and the risks of metastasis and local progression. The
absolute reduction in the risk of death after 10years is small, but the
reductions in the risks of metastasis and local tumor progression are
substantial 6.
In place of leaving people in blissful ignorance one can suggest them
bliss with awareness. Age old established practices like yoga have proven
efficacy of reducing stress and creating bliss from within 7, 8, and 9.
Reference:
1. Clinical utility of the percentage of positive prostate biopsies
in predicting prostate cancer-specific and overall survival after
radiotherapy for patients with localized prostate cancer. D'Amico AV,
Keshaviah A, Manola J, Cote K, Loffredo M, Iskrzytzky O, Renshaw AA.
Future Oncol. 2009 Dec; 5(10):1555-84.
2. Critical review of prostate cancer predictive tools. Shariat SF,
Kattan MW, Vickers AJ, Karakiewicz PI, Scardino PT. Curr Opin Urol. 2008
May; 18(3):279-96
3. Systematic Review: Comparative Effectiveness and Harms of
Treatments for Clinically Localized Prostate Cancer. Timothy J. Wilt,
Roderick MacDonald, Indulis Rutks. et al. Annals of Internal Medicine
2008; 148 (6), 435-448
4. Predictive models before and after radical prostatectomy.
Capitanio U, Briganti A, Gallina A, Suardi N, Karakiewicz The Prostate
2010: 70(12), 1371-1378
5. Bill-Axelson, A., et al., Radical Prostatectomy Versus Watchful
Waiting in Localized Prostate Cancer: the Scandinavian Prostate Cancer
Group-4 Randomized Trial. J. Natl. Cancer Inst., 2008. 100(16): p. 1144-
1154
6. Anna Bill-Axelson, Lars Holmberg, Mirja Ruutu, Michael Haggman,
Swen-Olof Andersson, Stefan Bratell, Anders Spangberg,Radical
Prostatectomy versus Watchful Waiting in Early Prostate Cancer. N Engl J
Med 2005;352:1977-84.
7. N. Janakiramaiah, B. N. Gangadhar, P. J. Naga Venkatesha Murthy
Antidepressant efficacy of Sudarshan Kriya Yoga (SKY) in melancholia: a
randomized comparison with electroconvulsive therapy (ECT) and imipramine
Journal of Affective Disorders 2000;57(1-3), 255-259
8. Richard P. Brown, Patricia L. Gerbarg. Sudarshan Kriya Yogic
Breathing in the Treatment of Stress, Anxiety, and Depression: Part II--
Clinical Applications and Guidelines. The Journal of Alternative and
Complementary Medicine. 2005, 11(4): 711-717.
9. Julienne E. Bower, Alison Woolery Yoga for Cancer Patients and
Survivors. Cancer Control , July 2005, Vol. 12, No. 3
Competing interests: No competing interests
Dear Neeru,
Thanks for your reply.
I , a GP and a trainer of GPs, would also like to believe that early
detection of Prostate Cancer would lead to improved treatment, and hence
survival. But neither your statistics, nor Sandblom et al's paper provide
any evidence at all for this belief - a triumph of hope over experience !
Wilson -Jungh Criteria for a good screening program are:-
the condition should be an important health problem
the natural history of the condition should be understood
there should be a recognisable latent or early symptomatic stage
there should be a test that is easy to perform and interpret,
acceptable, accurate, reliable, sensitive and specific
there should be an accepted treatment recognised for the disease
treatment should be more effective if started early
there should be a policy on who should be treated
diagnosis and treatment should be cost-effective
case-finding should be a continuous process
Unfortunately, as I and other Rapid Responders have pointed out, the
screened group in fact does worse than if they had been left in blissfull
ignorance , viz
"overall mortality in the screening group is 69/1492 (4.6%) versus
252/7487 (3.4%) in the control group"
Sandblom et al had stated their "Objective To assess whether
screening for prostate cancer reduces prostate cancer specific mortality",
but in fact omitted this key measure from the paper, and instead discussed
the highly misleading 'case-fatality rate'. I have calculated the "
Prostate-cancer-specific mortality" , which clearly demonstrates that the
requirement that "treatment should be more effective if started early" is
simply not met. The answer to Sandblom's et al's question is a
resounding "NO". Indeed, I would argue, with others before me, that
research should urgently be conducted into why this form of screening
actually appears to hasten one's death.
Sadly, this most important key " Prostate-cancer-specific mortality"
( expressed as death rates for screened and control groups) was missing
from the original paper, as several other Rapid responders have pointed
out, and only added later in Sandblom's 'Author's Response'. Howsoever
one reworks the other statistics ( which will inevitably suffer from lead-
time bias, and 'overdiagnosis' of clinically irrelevant cancers ), any
screening program which results in a higher death rate for those screened
cannot in my humble opinion be justified.
Dr Sam Lewis
Competing interests: No competing interests
1st comment: Neera states "Eighty five cases (5.7%) of prostate
cancer were diagnosed in the screened group and 292 (3.9%) in the control
group"
This implies to me that the screened group consisted of1492 individuals,
and the control group numbered 7487 (adding up to a total of 8979 ,
somewhat less than the 'Registry' n of 9026 persons).
My submission: I also say the same. 5.7% is equal to 85/1492 in
screening group and 3.9% is 292/7487 in control group (which is much less
than screening group percentage. We do not compare the actual numbers, we
compare the rates/percentages only.
2nd Comment: Neera states " The prostate cancer specific mortality was
30/85 (35%) for men with prostate cancer diagnosed in the screening group
and 130/292 (45%) for men with prostate cancer diagnosed in the control
group."
,br>This implies that for prostate-cancer, the crude mortality rate in the
screened group was 30/1492 ( 2% ) and the crude mortality rate in the
control group was 130/7487 (1.75%), which represents a very different
picture from Neera Gupta's claim that "There is significant difference in
prostate cancer mortality in the study groups more being in control (45%)
and less being in screening group (35%)". Neera has completely neglected
the frequent observation that screen-detected cancers are often less
aggressive, and may remain entirely asymptomatic, harmless and undetected
if the screening effort were not made.
My Submission : This I have copied and pasted from the article.
Actually it is not prostate cancer specific mortality, instead it is case
fatality rate which is 35% in screened group and 45% in control group.
30/1492 (2%) is prostate cancer specific mortality in screened group and
130/7487 (1.75%) in control group. There is erroneous use of these terms
in both the article and by us. These are not statistically different.
Actually the case fatality tells us the gravity and severity of the
disease and is significantly different in two groups telling that in
scrrened group the cancer is detected in a less severe stage/ early stage.
3rd Comment: Neera states "The overall mortality for men with
prostate cancer was 69/85 (81%) in the screening group and 252/292 (86%)
in the control group."
Again, the bigger picture is very different - overall mortality in the
screening group is 69/1492 (4.6%) versus 252/7487 (3.4%) in the control
group.
My Submission: Mortality also can be confounded by so many factors,
e.g. presence of other diseases, age of the subject at the recruitment and
follow-ups. We can only comment on mortality if the two groups were
identical in all the confounders.
4th Comment: Neera states "Observation: The median cancer specific
survival was 201 months in the screened group and 133 months in the
control group. Comment: Clearly, survival is better in screening group
than control group."
Neera has neglected 'lead-time' bias, whereby successful screening can
detect cancers long before they would present clinically, but not
necessarily add any effective therapy. Thus patients with cancer can
appear to live longer, simply by adding the 'lead-time' between detection
and clinical presentation, resulting in apparent, not real prolongation of
life. Worse, such patients have that much longer to live with the
knowledge of cancer, instead of 'blissful ignorance'.
My submission: Though I am a public health personnel, I do not agree
that in initial stages the cancer will not have effective therapy and
hence improved survival.
The statistics I have picked up from the article by copying and pasting. I
still feel that cancers in initial stages when they are localized are
better candidates for therapy.
Competing interests: No competing interests
Professor Bland raises an important point. This study has been
described as a randomised trial in this and other publications over the
years, but we agree that it is clearer and more accurate to describe the
selection process as systematic rather than randomised.
It seems there may not be universal agreement on terminology,
however, since the term "systematic randomisation" is sometimes used to
describe this method. For example, the 2010 European Medicines Agency
guide to veterinary clinical trials says "The most common randomisation
methods are simple randomisation, systematic randomisation, stratified
randomisation and block randomisation."
[http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guidelin...
]
In any case, the approach used still results in a probability sample.
Our statistician comments that, "With Nth name selection and a pre-
existing list (as in this case), randomly selecting the 1st person from
between 1 and N in the list would give all individuals in the sample the
same chance of being selected, a pre-requisite for random selection. Thus
in this particular case the distinction between systematic and random
selection is a fine one, and the group allocated to screening is no more
likely to be biased than if traditional randomisation had been used."
We hope this matter will not detract from the message of the study,
which still holds. Many thanks to Martin Bland for spotting this.
Competing interests: Elizabeth Loder works for the BMJ and participated in decisions about this paper.
Neera Gupta, in her Rapid Response to Sandblom et al, makes wholly
unwarranted claims that Prostate Cancer Screening is beneficial, using
inappropriate statistics selectively.
Neera states "Eighty five cases (5.7%) of prostate cancer were
diagnosed in the screened group and 292 (3.9%) in the control group"
This implies to me that the screened group consisted of1492
individuals, and the control group numbered 7487 (adding up to a total of
8979 , somewhat less than the 'Registry' n of 9026 persons).
Neera states " The prostate cancer specific mortality was 30/85 (35%)
for men with prostate cancer diagnosed in the screening group and 130/292
(45%) for men with prostate cancer diagnosed in the control group."
This implies that for prostate-cancer, the crude mortality rate in the
screened group was 30/1492 ( 2% ) and the crude mortality rate in the
control group was 130/7487 (1.75%), which represents a very different
picture from Neera Gupta's claim that "There is significant difference in
prostate cancer mortality in the study groups more being in control (45%)
and less being in screening group (35%)". Neera has completely neglected
the frequent observation that screen-detected cancers are often less
aggressive, and may remain entirely asymptomatic, harmless and undetected
if the screening effort were not made.
Neera states "The overall mortality for men with prostate cancer was
69/85 (81%) in the screening group and 252/292 (86%) in the control
group."
Again, the bigger picture is very different - overall mortality in the
screening group is 69/1492 (4.6%) versus 252/7487 (3.4%) in the control
group
Neera states "Observation: The median cancer specific survival was 201
months in the screened group and 133 months in the control group.
Comment: Clearly, survival is better in screening group than control
group."
Neera has neglected 'lead-time' bias, whereby successful screening can
detect cancers long before they would present clinically, but not
necessarily add any effective therapy. Thus patients with cancer can
appear to live longer, simply by adding the 'lead-time' between detection
and clinical presentation, resulting in apparent, not real prolongation of
life. Worse, such patients have that much longer to live with the
knowledge of cancer, instead of 'blissful ignorance'.
Thus, it is clear to me that Prostate Screening is, in crude and
general terms, bad for my health. I dismiss Neera Gupta's unwarranted
spin, and agree with Sandblom's original conclusions :-
"After 20 years of follow-up the rate of death from prostate cancer
did not differ significantly between men in the screening group and those
in the control group."
Competing interests: No competing interests
Screening for prostate cancer should be advocated as it helps in early diagnosis for prostate cancer
The study on screening the prostate cancer is very interesting but
the results should be interpreted cautiously. We need to have a closer
look into the following observations made in the study1:
Observation: Eighty five cases (5.7%) of prostate cancer were
diagnosed in the screened group and 292 (3.9%) in the control group1.
Comment: This means that screening was significantly effective in
diagnosing the cancer.
Observation: In the intervention group, 43 tumours were found at
screening and 42 in the interval between examinations . The percentage of
men with localised tumours (T1-2, N0/NX, M0) was significantly higher in
the screened group (56.5%) than in the control group (26.7%, P<0.001).
The rates of non-localised tumours were 37/1494 (2.5%) in the screening
group and 213/7532 (2.8%) in the control group (P=0.44)1.
Comment: The localized tumour were diagnosed more on screening that
means the curable cancer (early stages) was diagnosed on screening which
would have higher survival rates.
Observation: The prostate cancer specific mortality was 30/85 (35%)
for men with prostate cancer diagnosed in the screening group and 130/292
(45%) for men with prostate cancer diagnosed in the control group. The
overall mortality for men with prostate cancer was 69/85 (81%) in the
screening group and 252/292 (86%) in the control group1.
Comment: There is significant difference in prostate cancer mortality
in the study groups more being in control (45%) and less being in
screening group (35%)
Observation: The median cancer specific survival was 201 months in
the screened group and 133 months in the control group1.
Comment: Clearly, survival is better in screening group than control
group.
Thus, overall diagnoses, early tumour detection, survival and case
specific mortalities show that screening is advisable.
Reference:
1. Gabriel Sandblom,Eberhard Varenhorst, Johan Rosell, Owe L?fman,
Per Carlsson. Randomised prostate cancer screening trial: 20 year follow-
up. BMJ 2011; 342:d1539
Competing interests: No competing interests
1. To Dr Burn, Ms Fraser and Mr Balasubramanian
1. The decision to exclude men older than 69 years was taken towards
the end of the study in order not to risk overdiagnosis in older men. Even
if this could have been done consistently from the start of the study, the
increasing age of the study population during the first three screening
rounds could not be considered to affect the outcome in a substantial way.
2. To investigate whether the detection rate and subsequent mortality
changed after the introduction of PSA as screening tool, we tested a model
with diagnosis prior to 1993 versus diagnosis after 1993 as a dichotomous
time-varying covariate. No significant difference in mortality was seen in
this analysis. Even if such analysis is not sufficient to assess
sensitivity of the screening tools, the lack of significant impact on
survival indicates that the screening tool does not contribute
substantially to the outcome.
3. The problem of false positive findings at the primary
investigation has been analysed in previous reports from the Norrkoping
screening trial [1].
4. The predictive values have not been addressed in the present
report. They are commented briefly in a previous study from the Norrk?ping
study [2].
5. Indeed, the figures show survival only of the men who were
diagnosed with cancer. Corresponding figures based on the whole cohort
would not show any deviation between the curves.
References
1. Pedersen KV, Carlsson P, Varenhorst E, Lofman O, Berglund K.
Screening for carcinoma of the prostate by digital rectal examination in a
randomly selected population. BMJ. 1990 Apr 21;300(6731):1041-4.
2. Varenhorst E, Berglund K, Lofman O, Pedersen K. Inter-observer
variation in assessment of the prostate by digital rectal examination. Br
J Urol. 1993 Aug;72(2):173-6.
2. To Dr Tolan
Mortality in the control group is in the denominator, i.e. prostate
cancer mortality was slightly higher in the screening group. The
adjustment for age was done a priori, we have not tested for any other
covariates no presented in the article. The analyses were done beginning
from January 1, 1987, not from the day of diagnosis. The hazard ratio 1.23
is without adjustment for age, whereas 1.58 is with age adjustment. Mean
age at start of the study was 61.25 in the screening group and 62.10 in
the control group (no significant difference).
We based the Cox proportional hazard analysis and Kaplan Meier
analysis on men with a diagnosis of prostate cancer. The adjustment for
age at diagnosis was made in order to minimise the lead time bias. The
main outcome measure, however, was the risk ratio for death from prostate
cancer, which is based on the whole cohort.
3. To Professor Bland
The way the group assignment was done, i.e. by allocating every sixth
man in the list to the screening, was done in a way that no confounding
could be introduced. We therefore believe that the study fulfils all
criteria of a randomised controlled study.
4. To Professor Schroeder
1. The number of men in the screening group is indeed much smaller
than in the ERSPC and PLCO studies, but the long follow-up compensates
gives it a statistical power to detect at major hypothetical differences
in prostate cancer mortality. All refusers were also included in the
analysis of prostate cancer death rate ratio.
2. The study protocol was not designed with mortality as primary
endpoint. Nevertheless, the long follow-up enables analyses with mortality
as outcome meaningful, albeit not with the same statistical power as the
ERSPC trial.
3. The large number of interval cancers (occurring also when PSA was
used as screening tool) undoubtedly has a large impact on the outcome. We
are sure that screening as it is performed nowadays is more effective.
4. Even if lead time bias cannot be taken into account in Kaplan
Meier projections, we still believe they provide important information.
In conclusion, we agree with most of the comments from Professor
Schroeder. The way the study was performed may, as pointed out in the
paper, have overestimated as well as underestimated the benefits from
prostate cancer screening.
5. To Dr Aronowitz
1. Most men attended at least two screening rounds (1268 men attended
at least once and 726 at least three times).
2. Radical prostatectomy and radiotherapy was practiced much more
sparsely when the study was performed than it is today. When performing a
study with a follow-up of more than two decades, the problem of aiming at
a moving target becomes inevitable. The routines today differ in many
aspects from those in the 1990's.
3. The high mortality in the screening group is probably explained by
the low percentage of men undergoing treatment with radical intent as well
as relapse following radical treatment.
6. To Dr Dreier, Ms Walter and Mr Meyer
We are well aware of the problem with lead time bias and
overdiagnosis. In order to avoid lead time bias, the Cox proportional
hazard analyses were based on time from the start of the study, not from
diagnosis. Overdiagnosis probably accounts for the statistically
significant HR seen in the Cox proportional hazard analysis with
adjustment for age. This analysis was, however, done with the screening
group as reference category, i.e. the mortality was lower in the screening
group than in the control group.
7. To Dr Junod, Mr Kaplan, Mr Foucras, Ms Dupagne and Mr Nicot
We used prostate cancer mortality in both groups as major endpoint,
i.e. including men without prostate cancer diagnosis in the denominators.
The risk ratio for death from prostate cancer determined in this was
estimated to (30/1494)/(130/7532)=1.16. As pointed out by Junod et al,
this risk ratio was not significant. Our intent was not show an
illusionary improved prognosis in the intervention group, but rather to
depict the long term survival of men diagnosed with prostate cancer in a
screening program, without neglecting the problem of overdiagnosis and
lead time bias.
Competing interests: No competing interests
Re: Randomised prostate cancer screening trial: 20 year follow-up
Excuse me if I have missed something, but is it true (as I have heard in a radio interview) that a large proportion of the so-called control group was voluntarily screening itself, individual by individual -- probably as a result of the publicity surrounding the study in the town concerned. Therefore the control group was not a valid control group, if what you wanted was a comparison of screened with unscreened men.
Competing interests: Men's Rights Activist