A schematic representation of true biological progression of the disease that can occur if initially low-grade cancer cells or histologically normal cells transform into higher-grade cancer cells
However, more importantly, reclassification is a function of diagnostic inaccuracy, i.e., undergrading of PC initially at diagnosis. Common diagnostic measures such as transrectal ultrasound and random 12-core biopsies frequently miss high-grade cancers as well as overdiagnose insignificant cancers as illustrated in Fig. 9.2 .
Radical prostatectomy specimen shown with multifocal disease and three theoretical routes for TRUS-guided transrectal biopsy needle causing undergrading. 1 Needle not hitting any cancer lesions (needle 1). 2 Needle hitting clinically insignificant low-grade cancer lesion (needle 2). 3 Needle hitting a tumor but missing high-grade part of it (needle 3)
Pre-PSA era PC was often a clinical diagnosis and based on findings in TURP specimen and referred to as “indolent” PC. Such patients were often not followed as strictly as patients in AS protocols today, and yet, excellent PC-specific survival has been reported in localized low-grade disease .
The PSA era is characterized by a dramatic increase in PC incidence throughout the Western world mainly due to widespread use of serum PSA for early diagnosis. PC diagnostics improved with the utilization of transrectal biopsies under ultrasound (US) guidance, replacing fine needle aspirations, and the systematic use of Gleason grading system for reporting biopsy results. Concern about potential overdiagnosis and overtreatment was soon raised . Thus, there have been efforts to define clinically insignificant PC, such as the Epstein criteria . These formed the basis for defining inclusion criteria and triggers for intervention in the subsequent AS cohorts and trials.
Several AS cohorts have been published, identified, and reviewed . Most of the AS protocols rely on repeat biopsies to monitor the disease. The volume of the disease is generally monitored with biopsy-based surrogates, e.g., number of positive biopsies and cancer length, and PSA-based surrogates, e.g., PSA, free PSA, PSA density (PSA-D), PSA doubling time (PSA-DT), and PSA velocity (PSA-V). However, the triggers used for reclassification/progression vary remarkably between the published cohorts [8–14]. Generally, treatment is recommended if during follow-up the patient no longer meets the inclusion criteria. A summary of the triggers used in some of the well-known AS cohorts is shown in Table 9.1.
Triggers for intervention in AS protocols
Triggers for intervention
GS upgrade, PSA-DT < 3 yearsa
GS >6, >2 positive cores, >50% core involvement
GS >6, >33% positive cores, >50% core involvement
GS >6, >2 positive cores, increase in core involvement
GS >3 + 4, >50% positive cores, PSA-V > 1
GS >6, >2 positive cores, PSA-DT < 3 yearsb
The initial reports of MRI ’s negative predictive value being close to 100% for clinically significant cancer seem promising in the AS setting . Also, there have been noteworthy reports concerning the initial use of genetic tests to predict cancer outcome . This makes it tempting to speculate that selection of patients for AS and triggers for intervention will largely rely on MRI, combined with targeted biopsies, and genetic biomarkers in the future. However, the value of using MRI and genetics for surveillance in patients with PC on AS still awaits confirmation in prospective trials with sufficiently long follow-up.
Here, we take a detailed historical perspective to review the literature on triggers for intervention in PC patients on AS.
Before PSA became a widely accepted tool for prostate cancer (PC) diagnostics and monitoring, PC was often an incidental finding in TURP performed for obstructive urinary symptoms due to benign prostatic hyperplasia or in cystoprostatectomy performed due to bladder cancer. These PCs were considered indolent and did not usually lead to PC treatments. Instead, conservative non-standardized follow-up, if any, was exercised. Biopsies, mostly fine needle aspirations in the early era, were not routinely repeated during follow-up, and if symptomatic disease developed, endocrine treatment was initiated. This protocol with deferred palliative treatment was commonly referred to as watchful waiting.
Despite poor diagnostic work-up and quality by today’s standards, as well as lack of regular follow-up and triggers in the pre-PSA era , excellent long-term survival has been reported. In a population-based study from Iceland, 100% cancer-specific survival was reported for patients with pT1a PC . Additionally, in two better-known and commonly cited studies, similar findings of excellent PC-specific survival were observed for local low-grade tumors in the absence of follow-up protocol and triggers for curative intervention [17, 18]. Therefore, it may be justified to question the value of protocol-based strict follow-up, as in published contemporary AS series, for low-grade, low-stage tumors, especially considering up to 10-year lead time in PC diagnosis induced by PSA screening . However, the key in AS is to correctly balance the risk of symptomatic PC against competing risks of death due to comorbidities and age, as illustrated. In this respect it is intriguing that life expectancy has increased significantly throughout the world since these famous “watchful waiting” cohorts were published. As an example, in Finland life expectancy for a man was 67.94 years in 1977, when the first patients entered Johansson’s pivotal study, while it was 78.17 in the year 2014 . While the more than 10-year increase in life expectancy may compensate for PSA-induced lead time, it also certainly emphasizes the importance of long-term follow-up of AS cohorts.
PSA was first approved by the FDA for disease follow-up after radical prostatectomy in 1986 . Soon its potential as a diagnostic tool was realized, and this use was also approved by the FDA in 1994 . The era is characterized by a massive increase in PC incidence due mainly to unorganized PSA-based screening, standardization of biopsies (TRUS-guided systematic biopsies), and redefined tumor grading (Gleason grading has replaced WHO grading). Two large, prospective PSA-based screening trials (ERSPC and PLCO) were launched, and it soon became evident that overdiagnoses and overtreatment were problems inherent in PSA-based diagnostics [22, 23]. This played a large role in the later recommendation of the United States Preventive Services Task Force (USPSTF) against the use of PSA screening. Eventually, efforts to differentiate clinically significant PC, i.e., those requiring treatment, from clinically insignificant PCs, i.e., those not requiring treatment, were initiated.
It is well accepted that candidates for AS are low-stage, low-grade tumors, while the roles of tumor volume and PSA in candidacy for AS are less clear. Despite a plethora of publications on “novel,” RNA- or DNA -based potential biomarkers for PC, currently none are widely accepted for clinical use. Instead, tumor volume and PSA-based triggers are included in most of the published AS cohorts and guidelines.
Stamey et al. laid the basis for determining clinically insignificant PC by evaluating the incidence of PC in a consecutive series of 139 cystoprostatectomies . They assumed that the pre-PSA era lifetime risk of PC (8%) would apply for the cystoprostatectomy cohort in which 55 PC index lesions were found. The number of the largest index lesions considered clinically significant stood at 11/55 (8%), clinically significant as being expected to be diagnosed during a man’s lifetime in the absence of PSA. The 11 index lesions were all over 0.5 cm3, which subsequently became the cutoff volume for clinically significant disease. In order to translate this to clinically usable criteria, as RP data is not available at diagnosis, Epstein et al. sought to look for diagnostic variables that could predict clinically insignificant versus significant PC as defined by Stamey. In a series of 157 consecutive RPs in patients with T1c disease, Epstein presented the following criteria: PSA density (PSA-D) ≤ 0.15, biopsy Gleason score ≤ 6, ≤ two positive biopsy cores, and ≤50% involvement of any biopsy core . These two studies laid the foundation for future research on AS for patients with PC.
However, the logic in the abovementioned studies and the conclusions thereof have also been criticized. In an attempt to repeat Stamey’s work in a more contemporary PSA era cystoprostatectomy cohort, Winkler et al. found 58 PCs (60%) in a series of 97 cystoprostatectomies. With the pre-PSA era assumption of 8% incidence for PC, as in the Stamey’s paper, the cutoff volume for significant PC would have been 1.09 cm3 . In a following study, Wolters et al. on the other hand looked at the Rotterdam ERSPC screening cohort and concluded that 1.3 cm3 for index tumor volume and 2.5 cm3 for total tumor volume were more appropriate cutoff values for low-grade and low-stage PCs . Importantly, all these studies merely establish general guidelines as the “one size fits all” approach does not exist in this context. The key is in the relation between tumor characteristics and competing risks of mortality, both of which are moving targets (slowly increasing aggressiveness of the tumor due to biologic progression versus shortening life expectancy due to comorbidities and increasing age during surveillance), as emphasized.
The first papers on AS cohorts were published in 2002. In the paper by Choo et al. the early experiences from the Toronto cohort were published . The authors state that the triggers used for intervention, namely, clinical, histological, and PSA progression, were arbitrarily defined, but it was concluded that PSA-DT may reflect tumor growth and predict its biological behavior. On the contrary, preliminary results of an AS cohort in the Johns Hopkins showed that PSA velocity did not correlate with disease progression, while PSA-D and free PSA did correlate . They used the aforementioned triggers for progression during surveillance, which was based on yearly rebiopsies.
The rationale for using PSA kinetics as a tool to monitor PC is intuitive and was first supported by a paper in which PC tumor volume was shown to correlate with serum total PSA value . However, it was later realized that over time the correlation between tumor volume and PSA has diminishes dramatically, likely due to decreasing tumor volume versus prostate volume ratio during the PSA era .
At the conclusion of the “early” PSA era, a candidate patient for AS was defined, triggers for intervention started to evolve, and the role of PSA kinetics was emphasized.
Characteristics of the “later” PSA era included the refinement of the Gleason grading system, continuing controversy over PSA screening, lack of support for the immediate curative treatment of low-risk PC [31, 32], start of several AS cohorts, publications of short-term AS results, and rise of MRI as an imaging tool in localized PC. Additionally, AS was finally accepted as a treatment option in most of the guidelines .
Gleason grading has evolved significantly over the years. Importantly, after the ISUP consensus meeting in 2005, only some of the cribriform pattern glands, those corresponding to the surrounding benign glands in size, could be considered as Gleason grade pattern 3 . Ultimately, from 2010 onward the view that any size PC with cribriform architecture should be considered as Gleason grade pattern 4 has been widely adopted [35–37]. Notably, it has also been recognized that no Gleason score of 2–4 should be made on needle biopsies, a concept that had already earlier been proposed by some authors . These changes in the diagnosis of Gleason scores 6 and 7 defined the so-called modified Gleason score and have resulted in disease upgrading. In other terms, a Will Rogers phenomenon has occurred in which the average aggressiveness of both Gleason score 6 and 7 subgroup cancers has decreased. Thus, both Gleason 3 + 3 and 3 + 4 PC in a surveillance biopsy today are likely to be associated with a better prognosis than they were before 2005 and 2010.
Currently, there is no evidence in the literature that definitive treatment of Gleason 6 PC prolongs survival. In fact, there is very little evidence that Gleason 6 cancer behaves like a cancer at all despite having the required histological features . Also, it is difficult to find evidence from the literature for Gleason 6 PC to spread to lymph nodes, distally or to cause mortality [39, 40]. Thus, the current thinking is that Gleason 6 itself does not pose a threat to patient but is merely a risk factor for higher-risk disease. The clinical implication is that volume of Gleason 6 should not be used as the sole trigger for intervention but as a trigger for further tests to exclude co-existent higher-risk disease.
How much, if any, Gleason 3 + 4 is allowed initially or at surveillance biopsies for a man considered for AS or on AS? There is no randomized data showing survival benefit for these cancers with immediate curative treatment . There is registry data suggesting that treatment consisting of a radical prostatectomy, delayed until a median of 19 months after diagnosis, did not affect the treatment outcome . There is no “one size fits all” solution. The decision must be based on an individual risk assessment and is highly influenced by a patient’s perception of treatment-related harms versus potential disease-related risks. Interestingly, in a recent autopsy study up to 50% of the cancers in unscreened Japanese men over 70 years of age were of Gleason 3 + 4 . This suggests that especially among elderly men, Gleason 3 + 4 PC may pose only a low risk and thus be a potential candidate for AS.
PSA Kinetics Refined
Despite the fact that virtually all AS protocols incorporate PSA kinetics in their follow-up, the results are conflicting. In the PRIAS trial, PSA-DT was the second most common reason to trigger radical prostatectomy. However, in patients with PSA-DT as the sole trigger for treatment (radical prostatectomy), the surgical specimen showed favorable outcome (Gleason 3 + 3 and pT2) in almost half of the patients (46%) suggesting that PSA-DT is not specific enough (almost half the patients would undergo unnecessary treatment) . Also, in the accompanying regression analysis, PSA-DT did not predict adverse pathology on RP specimen. Similarly, a report from an AS cohort from UCSF concluded that PSA-DT did not correlate with biopsy progression . In the Johns Hopkins AS cohort in which treatment change is triggered only by adverse findings in yearly repeat biopsies, PSA-DT could not predict adverse rebiopsy findings while PSA-V was marginally significantly predictive (p = 0.06). However, no single, usable cutoff value could be identified in ROC analysis for either PSA-DT or PSA-V . Finally, Vickers et al. performed a systematic review of the literature and concluded that PSA dynamics is not a reliable trigger for treatment in early-stage PC . Challenge in using PSA kinetics as a trigger in AS lies in the fact that it is not specific but prone to significant variation due to infections, interventions, medications , and other causes as illustrated in Fig. 9.3, which describes PSA fluctuations during follow-up in one selected patient. At diagnosis, this patient had two cores positive for PC (0.8 mm and 0.4 mm cancer foci). He has had three systematic scheduled rebiopsies according to the PRIAS protocol with no cancer found. Six years after the diagnosis, he had a sudden increase in PSA that was explained by recently having been ill with flu, and his PSA soon dropped to its previous level. At that time, he had his first MRI with no sign of clinically significant disease. Subsequently, 10 years after diagnosis, the patient again had a sudden rise of PSA up to almost 30ug/L which was explained by a recent episode of urinary retention. An MRI was repeated and again with no sign of clinically significant disease (Fig. 9.3).
AS patient from PRIAS trial followed for 10 years and demonstrating fluctuation in PSA. This patient had two sudden rises of PSA explained by flu and urinary retention. No PC was detected in three follow-up biopsies or two multiparametric prostate MRIs
At the conclusion of the “later” PSA era, AS was well accepted as a treatment option for low-risk prostate cancer. Lack of prospective randomized or even long-term cohort data is evident. Several national and organizational guidelines and reviews have been published . In these, reclassification criteria vary remarkably (Table 9.1). Treatment is often recommended if during follow-up men no longer meet the entry criteria. Most of the guidelines recommend serial PSA measurements, DRE, and surveillance biopsies. However, controversy exists over the timing or interval of visits and the cutoffs used to trigger intervention. Generally, the value of PSA kinetics and monitoring Gleason 6 volume have been questioned as tools to trigger treatment.
Despite repeat surveillance biopsies being recommended in most of the guidelines and protocols , the adherence rate is poor. In the largest published AS cohort, the PRIAS trial , we recently showed that while adherence to PSA controls was excellent (91%), adherence to repeat biopsies decreased significantly over time (33% at 10 years) . This is likely to be caused by biopsy-related discomfort and complications . Among the most feared complications are septic infections although pain, hematuria, and hematospermia are also not uncommon . Furthermore, systematic, random, TRUS-guided biopsies initially miss around 30% of clinically significant cancers, while they overdetect clinically insignificant cancers (Fig. 9.2). Due to these systematic biopsy-related drawbacks, MRI has recently received a lot of attention. While the initial reports were not promising , the development of multiparametric MRI imaging techniques and structured reporting according to PI-RADS (Prostate Imaging Reporting and Data System) have revolutionized this use of MRI . Recently, a European School of Oncology Task Force recommendation for reporting MRI in men on AS was published . The aim is to standardize reporting and facilitate data collection. Ultimately, the aim is to develop criteria to distinguish true radiological disease progression during surveillance from natural variation.
Currently, the evidence for MRI comes mainly from its use in selecting patients for AS. Very high negative predictive values (NPVs), even up to 100%, have been reported for clinically significant disease [52–54]. In a recent retrospective analysis of 223 men eligible for PRIAS but undergoing immediate radical prostatectomy, the role of MRI in predicting upgrading was evaluated . In a multivariate model, typical clinical variables such as age, stage, PSA, PSA density, and number of positive cores were added in addition to MRI features (PI-RADS score). The PI-RADS score was the only significant predictor, with OR of 2.72 for every unit increase in PI-RADS score . According to a recent systematic review, two-thirds of patients suitable for AS have positive MRI . Positive MRI was more likely associated with upgrading in subsequent RP than negative MRI, while this was not the case for upstaging.