PSA screening recs from USPSTF

​In my humble attempt to go where angels rightfully fear to tread, I will review the new recommendations of PSA screening by the USPSTF ​(see psa uspstf review jama2018

in dropbox, or doi:10.1001/jama.2018.3712 for their detailed evaluation, and psa uspstf review jama2018 or doi:10.1001/jama.2018.3710​ for the recommendations)

Background:

-- prostate cancer is the most commonly diagnosed cancer in men in the US and the 2nd leading cause of cancer death. In 2018 approximately 165,000 men will be diagnosed and 29,000 will die from prostate cancer.

-- The lifetime risk of being diagnosed with prostate cancer in the United States is 13% and the lifetime risk of dying is 2.5%

-- Prostate cancer incidence is 74% greater in African-American men vs white men, and the lifetime risk of prostate cancer death is 4.2% for African-American men, 2.9% for Hispanic men, 2.3% for white men, and 2.1% for Asian/Pacific Islanders
-- median age at death from prostate cancer is 80 years old.​

-- autopsy studies from men dying from other causes find that more than 20% of those age 50 to 59 and more than 33% of those age 70-79 are found to have prostate cancer, though typically well-differentiated

-- other major risk factors besides race include older age and family history of prostate cancer. Other weaker risk factors include diets high in fat and low in vegetable consumption, and cigarette smoking. there are also studies suggesting an association between decreased ejaculatory frequency and increased prostate cancer risk (eg, see prostate ca and ejaculatory frequency in dropbox, or Rider HR, Europ Urol. 2016; 70:974)

Details:

-- screening: the best data are for PSA-based screening; the evidence is insufficient to support other tests (eg, free PSA levels, genetic or adjunctive imaging tests). Digital rectal examination is not recommended because of lack of evidence.

-- Large RCTs done on PSA-based screening (all considered “fair quality”):  

    -- PLCO (the US Prostate, Lung, Colorectal, Ovarian screening program): 76,683 US men aged 55 to 74 were randomized to either annual PSA screening for 6 years or usual care, from 1993 to 2001. The threshold for intervention was a PSA of 4.0 ng/mL. This study was limited because approximately 46% of the control group received routine PSA screening from community physicians each year compared with 85% in the screening group (ie, a pretty contaminated study). No difference was found between groups in terms of death from prostate cancer after almost 15 years of follow-up

    -- ERSPC (European Randomized Study of Screening for Prostate Cancer): 181,999 men aged 50 to 74, from 7 European centers, from 1993 to 2003. Screening interval was 4 years at all sites except Sweden which used a 2-year interval. The threshold for biopsy was most commonly 3.0 ng/mL, with some sites using 4.0 ng/mL. Contamination was not systematically assessed, though 20% in the control group in the Netherlands had PSA screening. This trial did find a positive effect from screening: number-needed-to-screen was 781 men aged 55 to 69 to prevent one man from dying of prostate cancer after 13 years. Some variability in the results from different sites, but prostate cancer mortality was significantly reduced in the Netherlands and Sweden; for example there was no significant benefit in the largest site (Finland) for prostate cancer mortality, but in Sweden there was an absolute risk reduction of 0.72%, a 42% relative reduction. In addition, 4 of the sites reported data on metastases, finding that after 12 years of follow-up PSA screening was associated with a 30% lower metastatic prostate cancer risk in those screened (absolute risk reduction 7 per 1000 men in the​ screening group vs 10 per 1000 men in the control group), an absolute reduction in the long-term risk of 3.1 cases per 1000 men screened

    --CAP (Cluster Randomized Trial of PSA Testing for Prostate Cancer): 408,824 men aged 50 to 69 in the UK were invited to a single PSA screening, 34% did so. PSA cutpoint was 3.0, follow-up 10 years, no difference in prostate cancer were all-cause mortality

-- harms from screening:

    -- false-positive tests: in ERSPC (most with cutpoint PSA of 3.0) 17.8% of men had at least one false positive result, and in Sweden where patients were screened every 2 years with a PSA cutpoint of 3.0 ng/mL, more than 45% had a false positive result over 10 years of screening. In PLCO, 10.4% of men received at least one false-positive PSA screen, with no cancer on subsequent biopsy. False-positive results were more common in ERSPC in the initial screening PSA (2/3 of tests were false positive). Older men had a higher false positive result

    -- harms of the procedures resulting from a positive screen include complications of prostate biopsy (pain, hematospermia, infection). Approximately 1% of prostate biopsies result in complications requiring hospitalization.

    -- Overdiagnosis: 20 to 50% of men diagnosed with prostate cancer by screening would never become symptomatic during their lifetime. For example, it is pretty striking that the ERSPC trial still showed that only 1 prostate cancer death would be prevented in 27 men diagnosed with prostate cancer by screening. [that's a lot of men potentially getting pretty aggressive treatment with its attendant morbidities to prevent one cancer death]

    -- psychological complications: within 3 cohorts (n= 1179), abnormal PSA screening followed by a normal biopsy resulted in increased prostate cancer specific worry up to a year after the biopsy, but no increase in depression or anxiety

    -- harms related to treatment, including erectile dysfunction, urinary incontinence, and bowel symptoms.

        -- Prostatectomy: 1 in 5 develop long-term urinary incontinence requiring the use of pads, 2 in 3 have long-term erectile dysfunction

        -- radiation therapy: half of men experience long-term sexual dysfunction, one in 6 develop long-term bowel symptoms including bowel urgency and fecal incontinence

Conclusions:

-- adequate evidence from RCTs suggest that PSA screening from age 55 to 69 may prevent 1.3 deaths from prostate cancer over approximately 13 years per 1000 men screened (as a perspective, this is not a huge benefit given how common prostate cancer is.  sort of like the not-so-great benefit of mammography screening in women: see http://gmodestmedblogs.blogspot.com/2014/04/mammograms-again.html 

-- screening programs may also prevent about 3 cases of metastatic prostate cancer per 1000 screened

-- the results from the above studies show no reductions in all-cause mortality from screening

Recommendations by USPSTF:

-- for men aged 55 to 69, the decision to screen should be on an individual basis; patients should have an opportunity to discuss the potential benefits and harms of screening with the clinician. Harms are greater for men >70 years old, and there is a recommendation not to routinely screen unless the patient expresses a preference for screening.

-- At this point, the PSA should be the test used, positive results may be followed with transrectal ultrasound-guided core-needle biopsy.

Commentary:

-- the USPSTF presumably gave the PSA screening studies only a "fair" quality rating because the PLCO study had lots of contamination (lots of the control patients had PSA done, biasing the results away from potential benefit), the single screen in the CAP study was still only done in a minority of enrollees, and even the ERSPC study (the one showing benefit of screening) did not show benefit at all sites, there were differences in recruitment methods and use of ancillary testing, the interval of screening varied from 2 to 4 years, the PSA cutpoint was more often 3.0 ng/mL though some sites used 4.0.

-- the ERSPC trial still showed that only 1 prostate cancer death would be prevented in 27 men diagnosed with prostate cancer by screening.

-- An analysis from the Prostate Cancer Prevention Trial: in the placebo group (ie not on finasteride), they found 2950 men who never had a PSA >4.0 mg/mL or abnormal digital rectal exam, yet (somehow) consented to have a prostate biopsy. AND 449 men (15.2%) had a diagnosis of prostate cancer, AND 67 of them (14.9% of these men) had a Gleason score of >7!!!!! In fact, the prevalence of prostate cancer was 6.6% in men who had a PSA <0.5 ng/ml. AND 4 of these 32 men had “high-grade” prostate cancer with Gleason score at least 7 (1 of them with Gleason score of 8, as well as another with Gleason 8 with PSA in  0.5 to 1 range). (see psa low and cancer nejm2004 in dropbox, or Thompson IM.Engl J Med 2004;350:2239-46.).  This rather striking study suggested lack of sensitivity of PSA screening even at remarkably low levels, though was not mentioned in the USPSTF review.

-- it should also be noted that the PSA also does not have great specificity: PSA can increase dramatically (>50 ng/mL) in patients with even localized prostatitis, and less significant increases occur with BPH (on the order of 0.1 ng/mL per gram of prostate tissue noted on ultrasound)​

-- and the data on prostate cancer picked up on screening mostly shows that it takes a really long time to show benefit:

    --the ProtecT trial in the UK of 1643 patients with Gleason score of 6 in 77%, 7 in 21% (i.e. patients mostly in moderate risk category), with mean PSA of 4.6, followed 10 years found: no difference in cancer specific survival rates (all 99%); the development of metastases (bone, lymph node, visceral, or serum PSA >100) had a slight difference favoring surgery of 2.4 vs 3.0 per 1000 person-years, though clinical progression was found in 22.9 per 1000 person-years in active monitoring vs 9 per 1000 person-years with aggressive therapy

    ​--PIVOT study: http://gmodestmedblogs.blogspot.com/2017/08/prostatectomy-not-help-for-localized-ca.html  , a 19.5 year followup of the PIVOT trail in men with localized prostate cancer, found that death from prostate cancer was quite low (5% assigned to surgery and 6% assigned to observation), but that men assigned to surgery had significantly less disease progression (41% vs 68%), or treatment for disease progression (34% vs 60%). local cancer progression occurred in 34% vs 62%, regional progression in 9% vs 14%, and systemic progression in 10% vs 15%. There was essentially no overall difference in deaths from prostate cancer until 12 years after the trial started. Those patients with initial PSA <10 did not show benefit til the 12 year analysis (the 4 and 8 year analyses had no difference in prostate cancer mortality between those on observation vs radical prostatectomy); those with initial PSA >10 did show a difference at the 8-year mark, with cumulative prostate-cancer mortality of 7.2% vs 2.4%

--ultimately, it is not so surprising that if one removes lots of prostates from lots of men that the incidence of prostate cancer might  decrease, BUT

    --for many men, this cancer is slow-growing and the benefit might take decades, as noted above. the adverse effects of aggressive therapies are immediate.

    --there have been articles I have seen in the past finding that there is equivalent mortality with prostatectomy vs not in those with more aggressive cancer, such as those who have larger increases of PSA over a few years (ie, not so clear that screening really helps those with aggressive prostate cancer)

    --and the PSA really is not a great test:

        ​--there are lots of overdiagnosis: up to 50% of men, by mathematical modeling, have documented prostate cancer, have (or would be recommended to get) prostate surgery or other intervention, yet never would die from prostate cancer.  Which translates to lot of men being incontinent, impotent, having bowel symptoms without any benefit. And the PSA does not have great specificity (BPH, prostatitis as above)

        --and, as noted above, there are men who have high grade prostate cancer with PSA <0.5 ng/mL !!

-- even in the only study with a documented benefit for screening (ERSPC), 781 men aged 55 to 69 need to be screened to prevent one man from dying of prostate cancer after 13 years​, and only 1 in 27 (<4%) of those diagnosed with prostate cancer on biopsy actually died from prostate cancer

So, how does one put this together (and what is a busy primary care provider to do)??

--we do need to find out the patient’s values. 

    --Are they so cancer-phobic that they want aggressive therapy at any cost??  In those patients it certainly makes sense to offer PSA screening, until a better test comes along. they should understand that treating a cancer, if found, does have the immediate possibility of adverse events with interventions, yet often a prolonged time to show benefit in prostate-related mortality. (and, if getting a PSA, make sure that the patient has not ejaculated in the past 2-3 days, since that can lead to false elevations)

    --are they so concerned about their current quality of life that they really would not want intervention in any case? Those patients should not be getting PSA screening

    --then, there is the group in the middle. It may be reasonable to continue brief discussions at subsequent visits to reassess where they are at, along with their understanding that if they  were in the minority to benefit from a pretty aggressive intervention, they might not realize it for a couple of decades

-- as with any of these tests, the sensitivity and specificity vary considerably by the cutpoint: we traditionally use a PSA cutpoint 4.0 in the US. It should be noted that in the European study (ERSPC) most centers use the cutpoint of 3.0, which would increase the sensitivity but signficantly decrease the specificity of the test (and increase the number of people getting biopsied by a lot).

--and, the reconsideration/elevation of the USPSTF recommendation is not a ringing endorsement of PSA screening: moving from a grade D (recommending against screening, moderate to high certainty that screening has no benefit or that harms outweigh benefits) to grade C recommendation (recommending selectively offering screening to individual patients based on professional judgment in patient preferences, moderate certainty that the net benefit small)