Redesigning prostate cancer screening strategies to reduce overdiagnosis

Introduction

It is remarkable that even though 30 years have passed since the introduction of prostate-specific antigen (PSA) into US clinical practice, we are still debating its value for prostate cancer screening. The US Preventive Services Task Force (USPSTF) has see-sawed from stating that PSA testing was a personal decision for the individual man (2008), to concluding “moderate or high certainty that … the harms outweigh the benefits” (2012) to recommending shared decision-making again (2018)(1). In most other industrialized countries, government-sanctioned guideline groups recommend against population-based screening, yet PSA testing remains common in everyday practice.

My own view is that the PSA debate has lost the forest for the trees. White papers and recommendation statements focus on subtle points of scientific methodology – such as whether post-randomization consent results in the loss of allocation concealment – while giving sparse attention to some larger facts and principles. In this opinion paper, I will lay out three simple, established and accepted propositions about PSA screening and argue that these lead to some straightforward conclusions

1. PSA screening has both harms and benefits

For many years, it was open to reasonable debate as to whether PSA screening does, in fact, reduce prostate cancer mortality. Recent data have ended this debate. There has been only one methodologically acceptable randomized trial of PSA screening, the ERSPC, and this clearly shows a reduction in prostate cancer mortality(2). Similarly, recent US population trends, with prostate cancer mortality reduced by 40–50% since the pre-PSA era, cannot be explained without reference to a screening effect(3).

The harms of PSA screening, on the other hand, are well-known. Prostate cancer generally has a very indolent course, such that a large proportion of screen-detected cancers would never have been diagnosed during the patient’s natural life. Overdiagnosis is a particular problem for prostate cancer because it is often coupled with overtreatment. Because of the proximity of the prostate to the bladder, penis and rectum, curative treatment often leads to long-term urinary, sexual and bowel symptoms.

2. As currently practiced, the harms and benefits of PSA screening are finely balanced

The ERSPC reported modest benefit to PSA screening(2), about a 20% reduction in the relative risk of prostate cancer death at 13 years, but a 50% increase in prostate cancer incidence. This leads to the widely cited figure that 27 men need to be diagnosed with prostate cancer to prevent one death. The fine balance between benefits and harms is no doubt why guideline groups such as the USPSTF recommend shared decision-making about PSA screening rather than recommending its routine adoption(1). The ratio of benefits to harms becomes even more questionable when we take into account how PSA is typically implemented in routine US practice, with screening most common in men over 70(4) and aggressive treatment rather than conservative management of most men with low-risk disease (5).

3. Extensive knowledge has developed on how to increase the benefits and decrease the harms of PSA screening

PSA screening is not a single intervention, for instance, 80 mg Aspirin for cardiovascular prevention, it can be implemented in many different ways. It is not hard to imagine programs that would almost certainly do more harm than good, screening only the very elderly, for instance, and treatment for indolent but not higher risk disease. Conversely, we can design approaches that positively shift the ratio of benefits to harms by using knowledge that has developed over the past decades. The five key elements – or as we call them: “the five golden rules” (6) – of an effective approach to PSA testing are the same as for any intervention: first, get consent; then don’t test, diagnose or treat patients unless it is clear that they will benefit from it, finally, use effective treatment.

Get consent.

Given the special issues with cancer screening – with consequences that can be life-changing or life-ending - no man should undergo PSA testing unless he himself expresses a wish to have the test and has discussed the pros and cons with his provider (shared decision-making).

Don’t screen men who won’t benefit.

There is clear evidence from the ERSPC that the benefit of PSA screening takes many years to accrue: survival curves only start to diverge 10 years after screening(2). As such, there is little benefit to screening men aged over 70. There is, however, considerable harm: approximately 40% of all overdiagnoses occur in this older age group(7). Furthermore, prostate cancer is a very slow growing disease, and there is compelling evidence that unless men have a higher than average PSA by the age of 60, they are highly unlikely to develop aggressive prostate cancer. In one well-known study, prostate cancer incidence and mortality was compared with men aged 60 receiving PSA screening and a cohort followed without screening, but who had given blood at age 60 as part of an epidemiologic study. In men with PSA less than the median, prostate cancer incidence was increased by screening, but prostate cancer mortality in the unscreened group was extremely low (~0.2% at 25 years) and was not reduced by screening(8).

Only biopsy men at elevated risk of high-grade disease.

PSA is a very specific marker for prostate cancer death(8), but has only moderate specificity. One common estimate is that for every 100 men with elevated PSA biopsied in typical US practice, 50 have no cancer, 30 have low-grade disease and only 20 have the sort of prostate cancer for which treatment is indicated(9). Recent years have seen the development of several reflex markers that can help inform the biopsy decision. Current guidelines recommend use of free-to-total ratio, the prostate health index, 4Kscore and PCA3(10). Use of such markers has been shown to reduce unnecessary biopsy and overdiagnosis of low-risk disease by approximately 50%, while still detecting most high-risk cancers.

Don’t treat men with low-risk prostate cancer.

At least 50% of men with screen-detected prostate cancer have low-risk disease. These men have a very low probability of prostate cancer mortality that is not importantly decreased by treatment(11). Naturally, such men will experience similar treatment-related morbidity from surgery or radiotherapy as men with intermediate or high-risk disease, for whom treatment is clearly indicated. They should be managed conservatively, an approach known as “active surveillance”.

If you have to treat, do so at a high-volume center.

There is clear evidence that both the harms and benefits of curative treatment for prostate cancer are highly dependent on how treatment is given(12). Far too many patients diagnosed with aggressive prostate cancer are given inadequate treatment. For instance, in 2005, the median yearly volume amongst surgeons who treated prostate cancer was 3(13).

Conclusion

A typical guideline group consider first whether PSA testing has any benefits, and if so, whether the benefits outweigh the harms. The problem is that the ratio of benefits to harms for PSA depends on how PSA testing is implemented. Guideline groups should first consider whether PSA testing has any benefits and, if so, how to implement PSA screening so that benefits are maximized and harms minimized. This needs to go beyond vanilla statements about shared decision-making and towards concrete recommendations as to who to screen, biopsy and treat.

Some back of the envelope calculations demonstrate just how large an effect some simple modifications to PSA screening can have. One estimate is that 42% of the overdiagnoses resulting from PSA occur in men aged 70 and over(7). If we then implemented routine use of markers in the remaining patients, with a 50% reduction in overdiagnoses, we would reach a total 70% reduction in overdiagnosis. Avoiding treatment in the remaining low-risk patients who are diagnosed would reduce overtreatment by greater than 70%. And, of course, by definition, reducing overdiagnosis and overtreatment does not reduce the effects of screening on mortality.

In sum, application of some basic medical principles – avoiding screening, biopsy or treatment in patients who are unlikely to benefit - coupled with recent findings on prostate cancer can transform PSA screening. Our question is therefore not whether to screen prostate cancer, but whether US medicine is prepared to make a few simple changes to ensure that we screen well.