Background
Currently 15% men are diagnosed with prostate cancer (PC) during their lifetime but the risk of death due to the disease is only 3% [
1]. The average expected years of life lost due to PC is 1.8 years, compared to other common malignancies such as breast cancer, at 16.7 years [
2]. To further emphasize this long natural history, up to 45% of patients are diagnosed with low risk PC [
3].
The rationale for active surveillance (AS) is that many patients have indolent tumors and would not benefit from immediate definitive treatment. However, there is always concern from both the patient and urologist that undiagnosed, aggressive disease exists. Indeed 30–50% of patients that meet AS criteria with Gleason Score 6 cancers are upgraded at radical prostatectomy with Gleason Score pattern 4 or higher being found [
4‐
8]. Subsequently, larger cohorts have shown that 10–38% of patients are reclassified on follow-up during AS due to Gleason Score upgrading or increased tumor volume (no. cores, maximal core involvement (MCI)) [
9‐
14]. It is imperative to improve the ability to identify those patients at higher risk for failing AS, either due to disease progression or reclassification, in order to provide better counseling regarding treatment options and to adjust surveillance schedules. In addition, it is not clear whether or not patients that fail AS do so as a result of finding previously undetected disease or disease progression.
The objective of the study was to analyze clinical and pathologic variables available at diagnosis to determine factors that could predict reclassification and AS failure. Our hypothesis is that having a biopsy showing bilateral tumor would indicate multifocal PC, which may be associated with higher volume and higher stage disease compared to those with unilateral tumors on initial biopsy.
Methods
Cohort definition
A retrospective review of a previously described protocol driven cohort [
15] of patients in the AS program at our institution was performed from 2007 to 2014. This was prior to the wider use of MRI for additional evaluation. Criteria utilized for AS were based on the National Comprehensive Cancer Network definition for very low risk disease including PSA < 10 ng/ml, Grade Group 1 (Gleason grade 6), ≤2 biopsy cores with cancer, and maximal core involvement (MCI) with cancer < 50%. PSA density was < 0.15 in 86% of patients, but density was not part of our exclusion criteria. Patients were followed with serial biopsies annually and delayed intervention using pre-identified clinical and pathologic cut-offs as trigger points to indicate disease progression. This study was approved by our Institutional Review Board.
Of 144 low risk patients in the AS program, 14 patients were excluded from analysis: 9 without a confirmatory biopsy, and 5 patients were T1a or T1b diagnosed after a transurethral resection of the prostate. In all patients a confirmatory biopsy was performed within 3–12 months after initial transrectal ultrasound guided (TRUS) biopsy was performed. Twelve to 16 core biopsies were obtained at each procedure. PSA was tested every 3–6 months per protocol and a follow-up biopsy was done within 1 year of diagnosis and then every 1–2 years. Patients who did not fail AS due to biopsy criteria, continued until they were reassigned to a watchful waiting (WW) non-intervention protocol due to advanced age or other significant comorbidities.
Patients were reclassified (e.g. failed AS criteria) and offered definitive therapy (radical prostatectomy or radiotherapy) if they met the following criteria on repeat biopsy: Grade Group upgrade, ≥3 cores positive for cancer, > 50% cancer core involvement or clinical progression defined as a palpable nodule greater than previously found or a change in the TNM classification.
We then performed a separate secondary analysis of 211 patients that met the same AS criteria over a similar time period (2007–2014) who declined AS and elected upfront radical prostatectomy. All pathology was read by a fellowship-trained urologic pathologist.
Statistical methods
Kaplan-Meier method was used to estimate time to reclassification (AS failure). Univariate and multivariate Cox proportional hazards regression analysis was performed to evaluate predictors of reclassification and AS failure. All analyses were performed using software SAS 9.4 (SAS institute Inc., Cary, NC, US). Matched pair analysis was performed for the radical prostatectomy and AS populations controlling for age, PSA < BMI, Prostate Volume, and PSA Density. Statisical support was provided by Dr. Glen Leverson, PhD.
Discussion
The selection of patients for AS according to current recommendations is guided by clinical and pathologic features that indicate the patient has a small, organ confined, well-differentiated tumor [
9‐
14]. Patients are labeled as having failed AS if they are found to have a pathological indication to suggest that the tumor is progressing (reclassification), including increasing tumor volume (increasing number of cores with cancer or increasing volume of cancer within a biopsy core) or Gleason score upgrading [
9‐
11,
13,
16]. Regular repeated biopsies play a pivotal role in redefining risk and reclassifying patients. Negative biopsies during surveillance follow-up occurs in 21–52% of patients and previous studies have shown these patients with no cancer on follow-up biopsies have a 53% reduction in risk of disease progression [
17,
18]. There are ongoing efforts to both improve the diagnostic yield of prostate biopsies and identify factors that may predict patients who are at higher risk for failing AS in order to provide better counseling regarding treatment options and avoid unnecessary interventions.
In our AS cohort, we found that patients with bilateral tumor on initial or surveillance biopsies were reclassified during AS follow-up over time earlier and more frequently compared to those with unilateral disease (66% vs 19.5% respectively). This risk increases over time (Fig.
2). The finding of bilateral disease has a higher rate of active surveillance failure than other clinical features including PSA density. PSA density was low (< 0.15) for the majority of the population making it a less powerful indicator of failure in this population. This finding is supported by a study in which the presence of bilateral prostate cancer on biopsy was exchanged with > 50% MCI as a reason for excluding AS excusion criteria, demonstrating good performace in predicting clinically significant prostate cancer [
19]. These findings suggest that the presence of bilateral disease is a clinically significant and inexpensive way to risk stratify patients enrolled in AS (Fig.
2). The presence of bilateral tumor can be used in counseling patients and as a trigger for further evaluation due to high failure rate in this group of patients.
In our analysis, bilateral disease on biopsy was the strongest parameter in predicting AS failure, outperforming other factors including the number of cores with cancer, MCI and PSA density. Other studies have assessed PSA kinetics to determine if this parameter can be used as a trigger for intervention. For example, in the PRIAS [
9] study, the largest ongoing prospective study, PSA doubling time < 3y was a trigger for intervention. However, Klotz et al. [
11] discontinued its use in 2009 and is not being used in other larger cohorts such as the John Hopkins AS group [
11] or UCSF. There is also suggestive data that PSA density can be a strong predictor for reclassification over longer term time periods, however more studies are needed to confirm these findings [
9,
20,
21]. In our data PSA density did not perform as well as the detection of bilateral cancer likely due to the fact that the majority (86%) of patients had a PSA density of < 0.15 and thus only a limited range was evaluated.
An important question is whether patients fail AS due to disease progression or due to detection at subsequent biopsy. Conceptually TRUS biopsy, although templated, is a procedure with a relatively low negative predictive value. In our separate analysis of a concurrent cohort of 211 patients eligible for AS but elected for upfront radical prostatectomy, 73% of these patients had bilateral, pT2c disease on final pathology. This was confirmed with our matched pair analysis. Studies have demonstrated that 71–80% of low risk patients have bilateral tumor on final specimen when taken for upfront radical prostatectomy [
22‐
24]. This suggests that the majority of the patients with PC have multifocal, bilateral tumors that are missed during TRUS biopsy even in a selected low risk population.
While it is important to fine tune the ability of urologists to use standard TRUS biopsy information to guide treatment counseling, MRI increasingly is being employed. In our active surveillance cohort, 42/130 (32%) patients had a prostate MRI either prior to diagnosis of prostate cancer, or as part of their follow-up. Recent improvements in MRI technology increase the yield of prostate biopsy, but not all practitioners worldwide have access to high quality imaging. To illustrate the limits of access to this technology, Japan has the most MRI units per million population (47) with the US second (38 upm). This is compared to countries like Germany (12 upm), France (11 upm), Canada (9 upm), UK (6 upm), and Mexico (2 upm) [
25]. MRI has a lower sensitivity of 63% in detecting lower volume (< 0.5 ml) and intermediate grade (Gleason 7) PC, but does have up to 80% detection rate for higher grade Gleason > 8 PC [
26]. There has been an increased use of MRI during AS to help determine eligibility and disease progression [
27]. These technologies are not yet widely adopted due to cost and access especially in underserved populations. MRI guided, targeted prostate biopsies may improve our management of AS patients and could be important in driving changes in biopsy schedules or AS criteria with additional data. With limited MRI access, these data demonstrating greater risk with bilateral positive biopsies may help direct those patients who need further radiologic evaluation.
Our AS failure rate is 38.5% is slightly higher compared to other cohorts as reported by Klotz et al. [
11], PRIAS [
9], Tosoian et al. [
10], Preston et al. [
12], Dall’Era et al. [
13], (22.6, 28, 30.6, 34.7, 38%). These variations in failure rates may result from different criteria or the population utilized for the study. Our study is limited in its retrospective nature and single institution, but it is strengthened by our assessment of a concurrent group of patients who met AS criteria and elected definitive treatment. A matched-pair analysis was performed in an effort to limit treatment bias in this population.