Cancer screening boondoggle?

[winstonfoot5]

Cant post something like this on twitter because /cancelled so posting here but does an ounce of prevention really equal a pound of cure? Docs and lay public alike often treat screening as prima facie

Discuss.

Estimated Lifetime Gained With Cancer Screening Tests

The meta-analysis examines evidence from randomized clinical trials of lifetime gained with use of screening tests for lung, breast, prostate, and colorectal cancer.

jamanetwork.com

JAMA Intern Med. 2023 Aug 28.
doi: 10.1001/jamainternmed.2023.3798. Online ahead of print.
Estimated Lifetime Gained With Cancer Screening Tests: A Meta-Analysis of Randomized Clinical Trials
Michael Bretthauer 1, Paulina Wieszczy 1 2, Magnus Løberg 1, Michal F Kaminski 1 2 3, Tarjei Fiskergård Werner 4, Lise M Helsingen 1, Yuichi Mori 1 5, Øyvind Holme 1, Hans-Olov Adami 1 6 7, Mette Kalager 1
PMID: 37639247 DOI: 10.1001/jamainternmed.2023.3798

Abstract
Importance: Cancer screening tests are promoted to save life by increasing longevity, but it is unknown whether people will live longer with commonly used cancer screening tests.

Objective: To estimate lifetime gained with cancer screening.

Data sources: A systematic review and meta-analysis was conducted of randomized clinical trials with more than 9 years of follow-up reporting all-cause mortality and estimated lifetime gained for 6 commonly used cancer screening tests, comparing screening with no screening. The analysis included the general population. MEDLINE and the Cochrane library databases were searched, and the last search was performed October 12, 2022.

Study selection: Mammography screening for breast cancer; colonoscopy, sigmoidoscopy, or fecal occult blood testing (FOBT) for colorectal cancer; computed tomography screening for lung cancer in smokers and former smokers; or prostate-specific antigen testing for prostate cancer.

Data extraction and synthesis: Searches and selection criteria followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guideline. Data were independently extracted by a single observer, and pooled analysis of clinical trials was used for analyses.

Main outcomes and measures: Life-years gained by screening was calculated as the difference in observed lifetime in the screening vs the no screening groups and computed absolute lifetime gained in days with 95% CIs for each screening test from meta-analyses or single randomized clinical trials.

Results: In total, 2 111 958 individuals enrolled in randomized clinical trials comparing screening with no screening using 6 different tests were eligible. Median follow-up was 10 years for computed tomography, prostate-specific antigen testing, and colonoscopy; 13 years for mammography; and 15 years for sigmoidoscopy and FOBT. The only screening test with a significant lifetime gain was sigmoidoscopy (110 days; 95% CI, 0-274 days). There was no significant difference following mammography (0 days: 95% CI, -190 to 237 days), prostate cancer screening (37 days; 95% CI, -37 to 73 days), colonoscopy (37 days; 95% CI, -146 to 146 days), FOBT screening every year or every other year (0 days; 95% CI, -70.7 to 70.7 days), and lung cancer screening (107 days; 95% CI, -286 days to 430 days).

Conclusions and relevance: The findings of this meta-analysis suggest that current evidence does not substantiate the claim that common cancer screening tests save lives by extending lifetime, except possibly for colorectal cancer screening with sigmoidoscopy.

---

Members don't see this ad.

 

[Chartreuse Wombat]

I am waiting to see whether the American Cancer Society responds. The moral is that widespread, population-based screening has small effects on cancer mortality (if any)

 

[OTN]

We've known this for awhile. The low-dose CT data in smokers is decent, however, despite what this trial says. That and FOBT have decent data. I'm also a believer in the "prostate cancer stages have advanced since we stopped checking PSA" population-based data, but I do understand that PSA screening hasn't shown any significant survival benefit.

 

[madchemist89]

This Is population level data. We all know that earlier detection results in less treatment and better outcomes. Treating an intermediate risk, localized prostate cancer is a lot better than positive nodes or bone mets. With all the antiandrogens available, you might not detect a survival difference, but quality of life is very different between the two. Early detection is the difference between cure and chronic illness.

 

[winstonfoot5]

Also interesting and relevant to the discussion is how there are companies with $B-dollar valuations (eg Illumina $8B acquisition of GRAIL) built on this premise but when you actually look at the data even for seemingly simple/common-sensical screening tests the evidence is often surprisingly shaky with questionable endpoints selection for the supporting trials

 

[Lamount]

Speaking only about lung cancer... I would say the conclusions of this study are misleading and should be interpreted with caution. The analysis of lung cancer screening here was under-powered. There final analysis only included 20,000 pts.

The national lung cancer screening trial, which randomized >50,000 to LDCT vs X-ray had similar incidences of lung cancer deaths LDCT and control arms (248 vs 309 per 100,000 person-years in NLST and 0.23 and 0.3 per 100 person years in this study), however the NLST was able to demonstrate a statistically significant improvement forLDCT in lung cancer deaths vs. X-ray (20% relative improvement p =0.004). Assuming a yearly CXR is not more of a significant risk for lung cancer than a LDCT, I think it is safe to conclude that the LDCT screening results in a significant improvement in lung cancer mortality.

Furthermore, it does not appear that the authors actually analyzed impact on cancer mortality just, all cause mortality (which strikes me as a fairly glaring omission) -correct me if I am wrong, but it seems like they just reported cancer mortality and then compared all-cause mortality. if they HAD reported changes in cancer mortality, they would have reported a similar 20% reduction... and I bet it would have also been SS.

A 20% reduction in cancer mortality isn't a "small effect" in my book...

Edited to tag @Chartreuse Wombat

 

[NotMattSpraker]

This Is population level data. We all know that earlier detection results in less treatment and better outcomes. Treating an intermediate risk, localized prostate cancer is a lot better than positive nodes or bone mets. With all the antiandrogens available, you might not detect a survival difference, but quality of life is very different between the two. Early detection is the difference between cure and chronic illness.

My views on this keep evolving but I tend to agree with you these days.

 

[Chartreuse Wombat]

Speaking only about lung cancer... I would say the conclusions of this study are misleading and should be interpreted with caution. The analysis of lung cancer screening here was under-powered. There final analysis only included 20,000 pts.

The national lung cancer screening trial, which randomized >50,000 to LDCT vs X-ray had similar incidences of lung cancer deaths LDCT and control arms (248 vs 309 per 100,000 person-years in NLST and 0.23 and 0.3 per 100 person years in this study), however the NLST was able to demonstrate a statistically significant improvement forLDCT in lung cancer deaths vs. X-ray (20% relative improvement p =0.004). Assuming a yearly CXR is not more of a significant risk for lung cancer than a LDCT, I think it is safe to conclude that the LDCT screening results in a significant improvement in lung cancer mortality.

Furthermore, it does not appear that the authors actually analyzed impact on cancer mortality just, all cause mortality (which strikes me as a fairly glaring omission) -correct me if I am wrong, but it seems like they just reported cancer mortality and then compared all-cause mortality. if they HAD reported changes in cancer mortality, they would have reported a similar 20% reduction... and I bet it would have also been SS.

A 20% reduction in cancer mortality isn't a "small effect" in my book...

Edited to tag @Chartreuse Wombat

This argument on what is the appropriate endpoint for a screening test (disease-specific mortality vs all-cause mortality) has been ongoing for more than three decades. You can choose a side and make an argument but just be clear what your endpoint is. Twenty percent reduction in cancer mortality with minimal effects on all cause mortality. Always tradeoffs and we must be clear about the endpoint. To say that cancer screening saves lives without an effect on all-cause mortality is (to coin a phrase) misinformation.

Dr Welch has been in the middle of these arguments for decades. His editorial in the same issue is worth reading

Testing Whether Cancer Screening Saves Lives

This Special Communication examines the feasibility of testing all-cause mortality for multicancer screening.

jamanetwork.com

 

[NotMattSpraker]

This argument on what is the appropriate endpoint for a screening test (disease-specific mortality vs all-cause mortality) has been ongoing for more than three decades. You can choose a side and make an argument but just be clear what your endpoint is. Twenty percent reduction in cancer mortality with minimal effects on all cause mortality. Always tradeoffs and we must be clear about the endpoint. To say that cancer screening saves lives without an effect on all-cause mortality is (to coin a phrase) misinformation.

Dr Welch has been in the middle of these arguments for decades. His editorial in the same issue is worth reading

Testing Whether Cancer Screening Saves Lives

This Special Communication examines the feasibility of testing all-cause mortality for multicancer screening.

jamanetwork.com

Good editorial. I love this debate. I have to admit that I found a lot of validity in arguments from people like Vinay Prasad, but philosophically I disagree more as time goes on. Maybe that's because I am aging

I think the harms of screening are overstated. You can't use a negative study to say screening "doesn't save lives" then turn around and imply that screening causes harms that shorten lives. Also, there is little data on specifically demonstrating the (hypothesized) benefits of screening in reducing the burden of therapy for cancer, and thus improving QoL.

I also find the cost arguments to be hypocritical in the US. We do many costly things with zero benefit. I don't understand deciding to cheap out on screening of all things. How about stop approving drugs with that offer no clinical benefit, like Enza for "active surveillance" or aducanumab for Alzheimer's.

Certainly, Dr. Welch has great points. I really liked the comments by Dahut in response to that study as well (ACS CSO).

I asked this question on Twitter and no one answered. I am reading a book by a doctor that makes a (epidemiology) data-driven recommendation for colonoscopy at 40, even in patients without family history.

Would you pay out of pocket for that?

For what it's worth, I am not.

 

[Lamount]

This argument on what is the appropriate endpoint for a screening test (disease-specific mortality vs all-cause mortality) has been ongoing for more than three decades. You can choose a side and make an argument but just be clear what your endpoint is. Twenty percent reduction in cancer mortality with minimal effects on all cause mortality. Always tradeoffs and we must be clear about the endpoint. To say that cancer screening saves lives without an effect on all-cause mortality is (to coin a phrase) misinformation.

Dr Welch has been in the middle of these arguments for decades. His editorial in the same issue is worth reading

Testing Whether Cancer Screening Saves Lives

This Special Communication examines the feasibility of testing all-cause mortality for multicancer screening.

jamanetwork.com

That's a fair point...

Again, specifically regarding lung cancer...
The NLST did demonstrate a significant reduction in all-cause mortality for LDCT vs. CXR of 6% (p=0.02), with more than half of the excess deaths in CXR being due to lung cancer. Would again argue that the current analysis is under-powered to make assertions about all-cause mortality in lung cancer, but given that the 20% reduction in lung cancer mortality parallels the NLST, I think one can safely surmise a similar benefit would be realized, given enough patients.

My personal take is that this analysis is that the authors of this present study are a little over the skis when they try to argue that we shouldn't say screening saves lives (at least for lung cancer)

 

[communitydoc13]

I think the paper is a little dangerous. There are two things that concern me. The comparison that they are making (to patients enrolled in control arms) and the endpoint (overall survival in highly variable populations).

For some context, I believe only 14% of cancer diagnoses in the US are screening detected. It is not clear to me that screening is driving the bulk of cost of cancer care.

The main thing they do here is bin all cause mortality (sticking to intention to treat cohorts) in control and screening groups for multiple randomized clinical trials.

Now, what are the control groups in these trials? They are patients fully engaged in the health system and committed enough for clinical trialists to have long term survival data on them. They are also submitting to routine healthcare interventions and presumably examinations. We also know that the compliance with not getting screened in these trials is not the best.

So the comparison is between screened and going to the doctor vs going to the doctor and intentionally not getting screened but getting routine health care and H&Ps and maybe being subjected to some screening.

Now when you factor in time horizon of ~10-15 years, age of typical diagnosed patient, competing risks, and most importantly natural variance in time of death as well as how that is related to reporting of time of death in these clinical trials, it will be almost impossible to demonstrate a statistically significant difference in mortality. The fact that there is the peculiar result of a positive result for sigmoidoscopy, a negative result for colonoscopy and a negative result for lung cancer screening tells you something. Look at the 95% CI on impact of screening on life expectancy for lung cancer screening. It's almost 2 years! This is reflecting the high all cause mortality and variance in life expectancy of this group of people in general.

The conclusion of this work really should be, "Patients subjected to cancer screening die with high variability of all causes. This means that measures such as cancer screening are unlikely to impact the distribution of deaths in time in a way that is statistically significant for any but the absolute largest cohorts (as you approach infinity)".

To say that cancer screening saves lives without an effect on all-cause mortality is (to coin a phrase) misinformation.

You can absolutely save lives but have a difficulty demonstrating the impact on all-cause mortality. Not misinformation. This happens all the time. Much of the best medicine does not impact all cause mortality for the whole population in a way that is easily demonstrable.

 

[NotMattSpraker]

That's a fair point...

Again, specifically regarding lung cancer...
The NLST did demonstrate a significant reduction in all-cause mortality for LDCT vs. CXR of 6% (p=0.02), with more than half of the excess deaths in CXR being due to lung cancer. Would again argue that the current analysis is under-powered to make assertions about all-cause mortality in lung cancer, but given that the 20% reduction in lung cancer mortality parallels the NLST, I think one can safely surmise a similar benefit would be realized, given enough patients.

My personal take is that this analysis is that the authors of this present study are a little over the skis when they try to argue that we shouldn't say screening saves lives (at least for lung cancer)

If, culturally, we treated lung cancer the way we treat breast cancer, screening rates would be through the roof.

 

[Chartreuse Wombat]

I think the paper is a little dangerous. There are two things that concern me. The comparison that they are making (to patients enrolled in control arms) and the endpoint (overall survival in highly variable populations).

For some context, I believe only 14% of cancer diagnoses in the US are screening detected. It is not clear to me that screening is driving the bulk of cost of cancer care.

The main thing they do here is bin all cause mortality (sticking to intention to treat cohorts) in control and screening groups for multiple randomized clinical trials.

Now, what are the control groups in these trials? They are patients fully engaged in the health system and committed enough for clinical trialists to have long term survival data on them. They are also submitting to routine healthcare interventions and presumably examinations. We also know that the compliance with not getting screened in these trials is not the best.

So the comparison is between screened and going to the doctor vs going to the doctor and intentionally not getting screened but getting routine health care and H&Ps and maybe being subjected to some screening.

Now when you factor in time horizon of ~10-15 years, age of typical diagnosed patient, competing risks, and most importantly natural variance in time of death as well as how that is related to reporting of time of death in these clinical trials, it will be almost impossible to demonstrate a statistically significant difference in mortality. The fact that there is the peculiar result of a positive result for sigmoidoscopy, a negative result for colonoscopy and a negative result for lung cancer screening tells you something. Look at the 95% CI on impact of screening on life expectancy for lung cancer screening. It's almost 2 years! This is reflecting the high all cause mortality and variance in life expectancy of this group of people in general.

The conclusion of this work really should be, "Patients subjected to cancer screening die with high variability of all causes. This means that measures such as cancer screening are unlikely to impact the distribution of deaths in time in a way that is statistically significant for any but the absolute largest cohorts (as you approach infinity)".


You can absolutely save lives but have a difficulty demonstrating the impact on all-cause mortality. Not misinformation. This happens all the time. Much of the best medicine does not impact all cause mortality for the whole population in a way that is easily demonstrable.

In my book the burden of proof is on whomever makes a claim. To support the claim that "Screening saves lives" you should provide evidence that all-cause mortality is improved. I agree that it may be difficult to obtain that evidence but if you don't have the evidence don't make the claim.

You should also not shift the burden to others to refute your claim

Russell's teapot - Wikipedia

en.wikipedia.org

 

[communitydoc13]

To support the claim that "Screening saves lives" you should provide evidence that all-cause mortality is improved.

I think this is an example of statistical thinking without considering priors.

Screening is tough because the cohort is the entire screened population, most of which will not demonstrate the condition being screened for. It is also degraded by effective treatment of symptomatic disease, futility of treatment etc.

But let's do a thought experiment. Let's imagine that screening detection is tied 1:1 to mortality for a given condition. It's a condition that is never symptomatic and never detected incidentally. When detected, it is always cured but it kills people 100% of the time when not treated.

Let's say that the condition is present in 1/1000 people, that the time to death is always within 10 years from detection based on our screening protocol and the likelihood of death from other causes is roughly 10% over the same interval. (These numbers are not that crazy when compared to the lung cancer screening trial and the mortality number within 10 years is low.)

Now let's imagine we knew these things, but wanted to design a trial to establish that all-cause mortality was improved over a 10 year interval when we did the screening.

What would be our accrual goal for standard 80% power and p value of 0.05? Almost 3 million people.

But, we knew the screening saved lives.

The lung cancer trial is overwhelming IMO.

I do agree that cancer care period is pretty low value in terms of global health outcomes and life expectancy. This doesn't mean that we don't save plenty of lives.

 

[Chartreuse Wombat]

I think this is an example of statistical thinking without considering priors.

Screening is tough because the cohort is the entire screened population, most of which will not demonstrate the condition being screened for. It is also degraded by effective treatment of symptomatic disease, futility of treatment etc.

But let's do a thought experiment. Let's imagine that screening detection is tied 1:1 to mortality for a given condition. It's a condition that is never symptomatic and never detected incidentally. When detected, it is always cured but it kills people 100% of the time when not treated.

Let's say that the condition is present in 1/1000 people, that the time to death is always within 10 years from detection based on our screening protocol and the likelihood of death from other causes is roughly 10% over the same interval. (These numbers are not that crazy when compared to the lung cancer screening trial and the mortality number within 10 years is low.)

Now let's imagine we knew these things, but wanted to design a trial to establish that all-cause mortality was improved over a 10 year interval when we did the screening.

What would be our accrual goal for standard 80% power and p value of 0.05? Almost 3 million people.

But, we knew the screening saved lives.

The lung cancer trial is overwhelming IMO.

I do agree that cancer care period is pretty low value in terms of global health outcomes and life expectancy. This doesn't mean that we don't save plenty of lives.

You are making my point. It is absolutely true that widespread, population screening requires millions of people to show an effect.

I agree that lung cancer will probably be different given its natural history and ability to put people into risk groups for screening based on environmental exposure (smoking) and improved treatment but the cancers that most commonly screened for are colorectal, breast and prostate where.

The answer is better screening where we focus on people that are a higher risk and screen for multiple cancer simultaneously. I am open to the possibility that better screening will save lives; improvements in screening technologies and screening for multiple cancer will greatly reduce the number of patients required to show an effect but we still need to do the studies. (Which was the point of the Welch editorial)

 

[communitydoc13]

The answer is better screening where we focus on people that are a higher risk and screen for multiple cancer simultaneously.

Yeah, we're on the same page. I was just concerned about this particular paper, which I think will be fodder for reflexive anti-screening sentiment. I also agree with @Lamount that in the context of the existing lung cancer screening trials, the paper's concluding claim is a little far, at least as far as lung cancer screening goes.

I also agree with @madchemist89 that other endpoints (including toxicity and duration of treatment) are pretty high value.

 

[Chartreuse Wombat]

I haven't really seen this paper discussed in the mass media but you are correct to be concerned that the takeaway is ALL SCREENING IS BAD. Very difficult to have nuanced messages in the current money for clicks environment.

I wouldn't go as far as the chemists assertion that "Early detection is the difference between cure and chronic illness". Cures can be worse than the underlying disease and overdiagnosis and overtreatment remain a problem.

 

[winstonfoot5]

Great discussion!

 

[Mandelin Rain]

Great discussion!

SDN still has moments of lucidity between all the misanthropy.

 

[Chartreuse Wombat]

Related to the discussion

Opinion | Not Everything We Call Cancer Should Be Called Cancer

Calling something cancer can lead to aggressive treatment even if the cancer in question is unlikely to cause problems.

www.nytimes.com

 

[Mandelin Rain]

Related to the discussion

Opinion | Not Everything We Call Cancer Should Be Called Cancer

Calling something cancer can lead to aggressive treatment even if the cancer in question is unlikely to cause problems.

www.nytimes.com

Newspeak at it's finest. Henceforth, carcinoma will be known as "joydiagnosis".

 

[Fpg1245]

Hey now you have evidence to cut spots for multiple specialties.

 

[NotMattSpraker]

Related to the discussion

Opinion | Not Everything We Call Cancer Should Be Called Cancer

Calling something cancer can lead to aggressive treatment even if the cancer in question is unlikely to cause problems.

www.nytimes.com

If you like podcasts, this was excellent. Same author Eggner.

 

[RickyScott]

I think this is an example of statistical thinking without considering priors.

Screening is tough because the cohort is the entire screened population, most of which will not demonstrate the condition being screened for. It is also degraded by effective treatment of symptomatic disease, futility of treatment etc.

But let's do a thought experiment. Let's imagine that screening detection is tied 1:1 to mortality for a given condition. It's a condition that is never symptomatic and never detected incidentally. When detected, it is always cured but it kills people 100% of the time when not treated.

Let's say that the condition is present in 1/1000 people, that the time to death is always within 10 years from detection based on our screening protocol and the likelihood of death from other causes is roughly 10% over the same interval. (These numbers are not that crazy when compared to the lung cancer screening trial and the mortality number within 10 years is low.)

Now let's imagine we knew these things, but wanted to design a trial to establish that all-cause mortality was improved over a 10 year interval when we did the screening.

What would be our accrual goal for standard 80% power and p value of 0.05? Almost 3 million people.

But, we knew the screening saved lives.

The lung cancer trial is overwhelming IMO.

I do agree that cancer care period is pretty low value in terms of global health outcomes and life expectancy. This doesn't mean that we don't save plenty of lives.

keep in mind that if mankind were to totally eliminate/cure cancer, on average lifespan would be increased by around 2 years. That means that if the USA were to eliminate cancer in the same way we did smallpox, we could match Puerto Rico or Algeria in life expectancy!

 

[DoctwoB]

GIGO:

Median follow-up was 10 years for computed tomography, prostate-specific antigen testing,

Don't have access to fulltext but can't dive into their rationale, but there is no reason for 10 year PSA follow up. We have 16 year follow from the better of the two PSA screening trials (the ERSPC) A 16-yr Follow-up of the European Randomized study of Screening for Prostate Cancer - PubMed. We have 22 year follow up from the most rigorous trial, the Goteborg PSA screening trial (lateral partially incorporated into ERSPC). Results from 22 years of Followup in the Göteborg Randomized Population-Based Prostate Cancer Screening Trial.

10 year data strongly suggests they are looking at the PLCO trial, in which there was more PSA testing done in the control arm then in the PSA arm, aka garbage trial.

 

[DoctwoB]

If you like podcasts, this was excellent. Same author Eggner.

Feel like this argument goes in circles

TLDR:
Pathologist: Gleason 6 disease is pathologically cancer.
Urologist: But it doesn't behave like cancer!
Pathologist: It is cancer.
Urologist: It doesn't metastasize!
Pathologist: It is cancer.

I get their perspective, unless we are going to pathologically redefine what cancer is, then we shouldn't get them to BS just so we can tell our patient's different. I also think the "patient's want treatment' because it's cancer argument is BS. Any doctor with any communication skills should be able to explain what gleason 6 cancer is and why it shouldn't be treated.

I've treated one patient with Gleason 6 in 4 years, and it was because it was in 11/12 cores, PSA was getting near intermediate risk range, and MSKCC nomograms gave him extremely high risk of ECE. Sure enough pathology was GG2 pT3b.

I

 

[DoctwoB]

We've known this for awhile. The low-dose CT data in smokers is decent, however, despite what this trial says. That and FOBT have decent data. I'm also a believer in the "prostate cancer stages have advanced since we stopped checking PSA" population-based data, but I do understand that PSA screening hasn't shown any significant survival benefit.

PSA screening definitely has a significant CSS benefit. No proven OS benefit for the reasons discussed above.

Best available data is the Goteborg trial with minimal PSA contamination of control arm and 22 year follow up.

"After 22 years, 1,528 men in the SG and 1,124 men in the CG had been diagnosed with PC. In total, 112 PC deaths occurred in the SG and 158 in the CG. Compared with the CG, the SG showed a PC incidence rate ratio (RR) of 1.42 (95% CI, 1.31–1.53) and a PC mortality RR of 0.71 (95% CI, 0.55–0.91). The 22-year cumulative PC mortality rate was 1.55% (95% CI, 1.29–1.86) in the SG and 2.13% (95% CI, 1.83–2.49) in the CG. Correction for nonattendance (Cuzick method) yielded a RR of PC mortality of 0.59 (95% CI, 0.43–0.80). Number needed to invite and number needed to diagnose was estimated to 221 and 9, respectively. "

Next best is the ERSPC: moderate control group contamination, variable based on country cohort and variable follow up.

The rate ratio of PCa mortality was 0.80 (95% confidence interval [CI] 0.72-0.89, p<0.001) at 16yr. The difference in absolute PCa mortality increased from 0.14% at 13yr to 0.18% at 16yr. The number of men needed to be invited for screening to prevent one PCa death was 570 at 16yr compared with 742 at 13yr. The number needed to diagnose was reduced to 18 from 26 at 13yr.

 

[NotMattSpraker]

Any doctor with any communication skills should be able to explain what gleason 6 cancer is and why it shouldn't be treated.

Haha, yeah. And yet.

Totally agree this is a purely philosophical discussion. I am entertained by that stuff.

 

[communitydoc13]

PSA screening definitely has a significant CSS benefit. No proven OS benefit for the reasons discussed above.

Best available data is the Goteborg trial with minimal PSA contamination of control arm and 22 year follow up.

"After 22 years, 1,528 men in the SG and 1,124 men in the CG had been diagnosed with PC. In total, 112 PC deaths occurred in the SG and 158 in the CG. Compared with the CG, the SG showed a PC incidence rate ratio (RR) of 1.42 (95% CI, 1.31–1.53) and a PC mortality RR of 0.71 (95% CI, 0.55–0.91). The 22-year cumulative PC mortality rate was 1.55% (95% CI, 1.29–1.86) in the SG and 2.13% (95% CI, 1.83–2.49) in the CG. Correction for nonattendance (Cuzick method) yielded a RR of PC mortality of 0.59 (95% CI, 0.43–0.80). Number needed to invite and number needed to diagnose was estimated to 221 and 9, respectively. "

Next best is the ERSPC: moderate control group contamination, variable based on country cohort and variable follow up.

The rate ratio of PCa mortality was 0.80 (95% confidence interval [CI] 0.72-0.89, p<0.001) at 16yr. The difference in absolute PCa mortality increased from 0.14% at 13yr to 0.18% at 16yr. The number of men needed to be invited for screening to prevent one PCa death was 570 at 16yr compared with 742 at 13yr. The number needed to diagnose was reduced to 18 from 26 at 13yr.

Admittedly, when I look at the Goteborg trial, several things stand out that worry me.

First, it's pretty remarkable that without screening, 70% of prostate cancers are detected anyway. Were Swedish docs doing DRE as SOC? US docs are no longer doing this routinely. I actually think DRE is a reasonable part of the PCP exam in the absence of PSA..

Second, the OS survival numbers trended opposite the PCa survival numbers. Of course, the null hypothesis is that this is rando, but they did get OS numbers at 14, 18 and 22 years with about a 1/3 of men overall dying. At no timepoint did the numbers favor the screening cohort. So, maybe there is some signal there.

What could cause an inverse OS if real? Unfortunately, either localized treatments contributing to death or ADT contributing to life! (Dem eunuchs live long).

All PCa survival benefit is manifested in the high risk and advanced cohorts (within these cohorts disease must have been less severe in the screened populations).

Honestly, if I were to take this data as a starting point for a rational PCa strategy it would be: screen, but don't treat until high risk. Very simple. Forget about all that Decipher BS and personalized medicine in PCa.

It would be a cost savings policy and I suspect the best for population based OS. Terrible for radonc business.

 

[ramsesthenice]

Lets rephrase this issue: do we treat populations or individuals? Hard right? We do both and that is where the challenge comes in. The question of does any diagnostic test or treatment save lives comes down to how you define save lives. If you can show that fewer people are dying of whatever condition you are treating or diagnosing (such as CSS) then on a very basic level the answer to the question is yes. But in the bigger picture, deciding the cost/value of that is the hard part.

My bias is that when a patient is in the room, I am treating a patient and it comes down to informed consent. I don't routinely recommend treating GGG1 prostate cancer. But if someone has crippling anxiety and is set on treatment, I don't think it is appropriate to decline primary radiation which is still considered an acceptable (although not preferred) option. I'll make damn sure they know they are statistically more likely to have a complication from radiation than die of prostate cancer. But if that doesn't sway them, that is their prerogative.

I use a similar analogy to redirect the difficult question of "what are my chances of beating this." My answer is along the lines of, I can take all of your information and give you averages but the reality is that for you, there are only 2 possible answers. 0% or 100%. At this point, the average answer would be in the range of X, but once we see how you respond to this treatment (or don't), that number is going to change. Lets start with the best treatment we have, and reassess when we know more. More often than not, this actually does seem to mentally help people cope with uncertainty and long odds without giving the impression of false hope.

 

[DoctwoB]

Lets rephrase this issue: do we treat populations or individuals? Hard right? We do both and that is where the challenge comes in. The question of does any diagnostic test or treatment save lives comes down to how you define save lives. If you can show that fewer people are dying of whatever condition you are treating or diagnosing (such as CSS) then on a very basic level the answer to the question is yes. But in the bigger picture, deciding the cost/value of that is the hard part.

My bias is that when a patient is in the room, I am treating a patient and it comes down to informed consent. I don't routinely recommend treating GGG1 prostate cancer. But if someone has crippling anxiety and is set on treatment, I don't think it is appropriate to decline primary radiation which is still considered an acceptable (although not preferred) option. I'll make damn sure they know they are statistically more likely to have a complication from radiation than die of prostate cancer. But if that doesn't sway them, that is their prerogative.

I use a similar analogy to redirect the difficult question of "what are my chances of beating this." My answer is along the lines of, I can take all of your information and give you averages but the reality is that for you, there are only 2 possible answers. 0% or 100%. At this point, the average answer would be in the range of X, but once we see how you respond to this treatment (or don't), that number is going to change. Lets start with the best treatment we have, and reassess when we know more. More often than not, this actually does seem to mentally help people cope with uncertainty and long odds without giving the impression of false hope.

I agree with most of your statement, other then "it is their prerogative"

It is their prerogative to seek a certain treatment. It is my prerogative, and more importantly my obligation to not provide potentially life altering treatment without medical benefit. Obviously there is a lot of gray area here (high volume GG1, significant GG1 volume plus symptoms of BPH, maybe GG1 with adverse genomics/FH profile) where treatment is reasonable, but the overall point stands.

 

[NotMattSpraker]

Lets rephrase this issue: do we treat populations or individuals? Hard right? We do both and that is where the challenge comes in.

I love this question from a philosophy of medicine standpoint and think about this often. We have to do both and I wish

I agree with most of your statement, other then "it is their prerogative"

It is their prerogative to seek a certain treatment. It is my prerogative, and more importantly my obligation to not provide potentially life altering treatment without medical benefit.

The fact that you are even thinking about this probably sets you apart from most oncologists that would routinely treat a GG1 prostate cancer. In practice, if you say no, they will find someone that will treat it. Possibly with protons. Hopefully at least someone that treats the whole gland.

I tend to agree with ramses. If you have a long discussion with them about why it shouldn't be treated but they convince you to treat them, they are likely in good hands. In my opinion (key word opinion), focusing only on OS or CSS is not a good strategy for an oncologist and blinds you to the concerns of many patients. There are psychological and other factors to consider.

Think of all the people on auto-pilot that spend 15 minutes with this guy and turn around and give him 45 fractions of proton therapy. I've never seen someone admit to doing this openly, but medicare is paying people every year to do it. The data are the data right?

Is there any data at all supporting even sending a genomic test on a GG1 prostate cancer? I worry these genomic tests are running way ahead of the data. Did people forget the reversal we had with Oncotype intermediate patients?

 

[communitydoc13]

I worry these genomic tests are running way ahead of the data.

I remain skeptical of these tests. Their value is only in circumstances where there is discordance between clinicopathological features and test results (and concordance is high, so this subgroup is small). I do not believe that the value of these tests in such a circumstance has ever been validated?

From the Michigan Collaborative Paper:

Among NCCN low risk patients, reclassification was observed in 20/58 (34%) of patients with lower Decipher scores compared to 3/8 (38%) with high Decipher scores. Among NCCN favorable intermediate risk patients, reclassification was observed in 11/43 (26%) of patients with lower Decipher scores compared to 2/7 (29%) with high Decipher scores.

A common scenario in my practice now is a 75 y/o + guy with GG2 disease who comes to me because urology got a Decipher and it's somewhat high. Patient already on ADT. Complete reframing of the patient from AS or observation vs XRT to reflexive ADT+XRT.

In the community, is the Decipher ever used to de-escalate therapy?

 

[evilbooyaa]

I would argue Decipher should primarily be used to consider de-escalation of therapy....

Like a UIR patint with low Decipher getting RT alone
or a FIR patient with low Decipher getting AS

We have no idea if Decipher high patients benefit from the additional therapies we subject them to if we're treating a Decipher score and not a patient... Prognostic, not predictive.

 

[ramsesthenice]

I would argue Decipher should primarily be used to consider de-escalation of therapy....

Like a UIR patint with low Decipher getting RT alone
or a FIR patient with low Decipher getting AS

We have no idea if Decipher high patients benefit from the additional therapies we subject them to if we're treating a Decipher score and not a patient... Prognostic, not predictive.

I am inclined to agree with you. Its not standard, so I don't do it yet, but I would love to be able to tell someone with 1-2 cores of GGG4 and no other risk factors that six months of ADT is probably enough. 2 years feels like overkill in these situations.

 

[DoctwoB]

I remain skeptical of these tests. Their value is only in circumstances where there is discordance between clinicopathological features and test results (and concordance is high, so this subgroup is small). I do not believe that the value of these tests in such a circumstance has ever been validated?

From the Michigan Collaborative Paper:

Among NCCN low risk patients, reclassification was observed in 20/58 (34%) of patients with lower Decipher scores compared to 3/8 (38%) with high Decipher scores. Among NCCN favorable intermediate risk patients, reclassification was observed in 11/43 (26%) of patients with lower Decipher scores compared to 2/7 (29%) with high Decipher scores.

A common scenario in my practice now is a 75 y/o + guy with GG2 disease who comes to me because urology got a Decipher and it's somewhat high. Patient already on ADT. Complete reframing of the patient from AS or observation vs XRT to reflexive ADT+XRT.

In the community, is the Decipher ever used to de-escalate therapy?

I use very rarely (<10% of my patients). Sometimes I feel pressured to do it because patients get second opinions from my local academic center and they get it on every patient, some patients feel they are getting better care when they get their fancy printouts.

The setting I see a bit of use is in the GG2 patient trying to feel better about surveillance (though the majority of the time it will be concordant with clinical factors and not make a big difference).