Image-based monitoring of targeted biopsy-proven prostate cancer on active surveillance: 11-year experience

Abstract

Purpose

To report our 11-year experience of Active Surveillance (AS) program focusing on modern transrectal ultrasound (TRUS)-based monitoring of targeted biopsy-proven cancer lesion.

Methods

Consecutive patients on AS, who had targeted biopsy-proven lesion followed by at least a repeat surveillance biopsy and three times TRUS monitoring of the identical visible lesion, were included. Doppler grade of blood flow signal within the lesion was classified from grade 0 to 3. Biopsy-proven progression was defined as upgrade of Gleason score or 25 % or greater increase in cancer core involvement.

Results

Fifty patients were included in this study. Clinical variables (median) included age (61 years), clinical stage (T1c, 42;T2, 8), PSA (4.6 ng/ml), and Gleason score (3 + 3, n = 41;3 + 4, n = 9). Of the 50 patients, 34 demonstrated pathological progression at a median follow-up of 4.4 years. In comparing between without (n = 16) and with (n = 34) pathological progression, there were significant differences in cancer core involvement at entry (p = 0.003), the major axis diameter (p = 0.001) and minor axis diameter (p = 0.001) of the visible lesion at entry, increase in the major axis diameter (p = 0.005) and minor axis diameter (p = 0.013), and upgrade of Doppler grade (p < 0.0001). In multivariate analysis for predicting pathological progression, the increase (≥25 %) in diameter of biopsy-proven lesion (hazard ratio, 15.314; p = 0.023) and upgrade of Doppler grade (hazard ratio, 37.409; p = 0.019) were significant risk factors.

Conclusions

Longitudinal monitoring of the TRUS-visible biopsy-proven cancer provides a new opportunity to perform per-lesion-based AS. The increase in diameter and upgrade of Doppler grade of the lesion were significant risk factors for biopsy-proven progression on AS.

Introduction

Active surveillance (AS) is a standard option for low-stage prostate cancer [1]; however, the standardized protocol is still evolving. The two currently used monitoring tools for patients undergoing AS, including PSA kinetics [2–5] and systematic biopsy-driven data [6–8], have limitations. PSA kinetics is not cancer specific, with fluctuation by inflammation and mechanical manipulation. Since current surveillance biopsy data are derived from random systematic biopsies, biopsy-based data (Gleason score, number of positive cores, percent core involved or cancer core length) are subject to considerable variation due to sampling errors between entry biopsy and surveillance biopsy. Therefore, random systematic biopsies lack reliability in predicting disease progression during AS [8].

The primary challenge for tissue-preservation management of prostate cancer is accurate and precise spatial mapping of the baseline cancer location and extent. Evolving modern imaging may facilitate cancer diagnosis by visualizing lesions that may have been missed by random image-blind biopsies. Image visibility of the known cancer with multi-parametric TRUS could have several advantages, including allowing precise targeted biopsy for better characterization of the cancer lesion, precise spatial mapping to facilitate re-visit targeting, and precise “per-lesion follow-up” [9]. Magnetic resonance imaging appears to have a high yield for predicting reclassification or results of confirmatory biopsy to determine candidates for AS at the time of entry [10–12].

Using modern TRUS technology, TRUS staging by an expert sonographer can more accurately predict microscopic extraprostatic extension compared with established nomograms when cancer is previously confirmed from a TRUS-visible lesion [13, 14]. Increasing numbers of studies reported that prostate cancers detected by modern TRUS-guided targeted biopsies are of higher grade and larger volume and are therefore more likely to be clinically significant [7, 9, 15, 16]. Monitoring of modern TRUS-visible targeted biopsy-proven cancer lesion may become a useful monitoring tool for patients undergoing AS. In the present study, we report our 11-year experience of AS program to evaluate the role for monitoring of targeted biopsy-proven image-visible lesions on AS for per-lesion-based AS.

Patients and methods

From February 1998 to November 2009, a total of 552 patients entered into AS in our institute were retrospectively analyzed after obtaining institutional review board approval. For study inclusion, patients had to be in low- or intermediate-risk group in D’Amico’s classification [17] and be considered suitable candidates for AS. Eligible candidates had to be between ages 40 and 80 years with less than 20 ng/mL of PSA level, and have a life expectancy of more than 10 years. Surveillance protocol included 3–6 monthly PSA, 6–12 monthly TRUS, and surveillance biopsy within 3 years or if indicated. The TRUS-guided prostate biopsy procedure included detailed gray scale/ Doppler TRUS whole-gland image surveillance (10–15 min) followed by sextant template plus image-directed biopsies of any suspicious lesions in gray scale and/or Doppler TRUS images. This allowed detailed spatial documentation of any biopsy-proven cancer. TRUS with power Doppler was performed using a Type EUB-6500 ultrasonography system (Hitachi Medical Systems America, Inc. Tarrytown, NJ, USA) equipped with an endocavity probe, type EUP-V53 W. In TRUS survey, digitalized gray scale and Doppler images were recorded in hard drive of ultrasound machine. The three-dimensional schematic mapping of the suspicious lesion with documentation of its contouring and location in both axial (apex, mid, and base) and bilateral sagittal views of the prostate were documented. In each lesion, both major and minor axes of the dimensions (mm) of the suspicious lesion were documented, and Doppler grade of blood flow signals of the suspicious lesion was classified from 0 to 3 based on the comparison with the normal anatomical blood supply (NABS) in the other unsuspicious parts of the prostate in the current study; (0: no): no flow, (1: low): present but weaker or less than the NABS, (2: moderate): equal or similar to the NABS, and (3: high): stronger or greater than the NABS.

The patients who had not been followed up by TRUS image at least three times (n = 321) or surveillance biopsy (n = 252), and stopped AS for non-pathological reasons (e.g., anxiety, PSA elevation) (n = 75) were excluded in our study. Surveillance repeat biopsies with detail TRUS survey were performed within 36 months from the entry and correlated with baseline biopsy and imaging studies. The systematic biopsy was performed transrectally with six to twelve cores depending on prostate volume, and TRUS-guided biopsies of suspicious lesions in gray scale and/or Doppler image with at least two biopsies for each targeted lesion. The end point of AS was biopsy-proven progression defined as upgrade of Gleason score or 25 % or greater increase in cancer core involvement in comparison with baseline biopsy data. The biopsy-proven index lesion of each patient was primarily defined as having the highest Gleason score, and secondarily as the greatest cancer core involvement.

All statistical analyses were performed using IBM SPSS Statistics version.19 (IBM, Armonk, NY, USA). The patients’ clinical parameters and outcome of AS were analyzed using Mann–Whitney U tests. The influence of each variable on progression was assessed using Cox’s multivariate logistic regression analysis. p values of <0.05 were considered to indicate statistically significant differences.

Results

Fifty patients were included in this study. Median age of the 50 patients was 61 years (range 42–77). Median PSA level was 4.6 ng/ml (range 1.1–15.8). Median PSA velocity was 0 ng/ml/year (range −3.5–3.9 ng/ml/year). Eight patients (16 %) used 5-alpha-reductase inhibitor over 1 year. Patients with Gleason score 3 + 3 and 3 + 4 were 41 (82 %) and nine (18 %), respectively. Patients with clinical stage of T1c and T2a were 42 (84 %) and eight (16 %), respectively. The median diameter of the lesion at entry in the major axis and minor axis was 11 mm (range 4–28) and 8 mm (range 4–25), respectively. Patients with Doppler grade of 0, 1, 2, and 3 were 21 (42 %), 18 (36 %), eight (16 %), and three (6 %), respectively.

Median follow-up was 4.4 years (range 2–11 years). Of the 50 patients, 34 patients (68 %) demonstrated pathological progression. In comparing the patients without (n = 16) and with (n = 34) progression, there were significant differences in % core of cancer (%) at entry (p = 0.003), the diameter of the visible lesion at entry in the major axis (p = 0.001) and minor axis (p = 0.001), increase rate of the diameter of the visible lesion during surveillance in the major axis (p = 0.005) and minor axis (p = 0.013), and upgrade of Doppler grade (p < 0.0001) (Table 1). Figure 1a shows the longitudinal TRUS monitoring of a case without biopsy-proven progression. Figure 1b shows the longitudinal TRUS monitoring of a case with progression.

Table 1.

Patient characteristics at the last follow-up

	No progression	Progression	p value
No. of patients (%)	16	34	—
Age, year (range)	60 (42–77)	65 (49–75)	0.2
Clinical stage: T1c/T2a, no.	15/1	27/7	0.2
PSA, ng/ml (range)	4.3 (1.1–7.9)	4.8 (1.1–15.8)	0.6
PSA velocity, ng/ml/year, median (range)	0 (−1.1–3.0)	− 1.0 (−3.5–3.9)	0.2
Gleason score at first diagnosis: 3 + 3 = 6/3 + 4 = 7, no. (%)	11 (69)/5 (31)	30 (88)/4 (12)	0.1
% core of cancer (%) (at entry)	7 (5–40)	21 (5–95)	0.003*
Risk group at start of AS: low/intermediate, no. (%)a	10 (63)/6 (37)	22 (65)/12 (35)	0.9
Patient with 5ARI, no. (%)	1 (6)	7 (21)	0.2
The diameter of biopsy-proven index lesion (at entry)
Major axis, mm (range)	13 (5–18)	11 (4–28)	0.2
Minor axis, mm (range)	9(4–16)	8 (4–25)	0.6
The diameter of biopsy-proven index lesion (at surveillance biopsy of last follow-up)
Major axis, mm (range)	12 (6–13)	14 (7–38)	0.001*
Minor axis, mm (range)	7 (2–10)	10 (5–20)	0.001*
Increase rate of the diameter of biopsy-proven index lesion (at surveillance biopsy of last follow-up)
Major axis, % (range)	0 (−50 to 70)	20 (−40 to 530)	0.005*
Minor axis, % (range)	− 10 (−60 to 50)	10 (−40 to 500)	0.013*
Change of classification of Doppler grade of biopsy-proven index lesion (blood flow), no. (%)
Upgrade/stable/downgrade	1 (6)/14 (88)/1 (6)	24 (70)/8 (24)/2 (6)	<0.0001*
Time of image follow-up, no. (range)	4 (2–5)	3 (2–6)	0.9
Follow-up time, months (range)	49 (24–132)	56 (24–123)	0.5

AS active surveillance, HEL hypo-echoic lesion

^*Statistically significant

^aD’Amico classification system for prostate cancer

Fig. 1.

a Biopsy-proven progression case b No biopsy-proven progression case

In univariate analyses for predicting progression in upgrade of Gleason grade, Gleason score at first diagnosis (p = 0.016), cancer core involvement (%) at entry (p= 0.005), increase (≥25 %) in major axis diameter _{(0.047), and the upgrade of Doppler grade (}p = 0.002) were significant predictors of progression. In multivariate analyses, the upgrade of Doppler grade (hazard ratio, 7.148; 95 % CI, 1.809–63.145; p = 0.037) was significant predictor of progression (Table 2).

Table 2.

Univariate and multivariate analyses to predict upgrade of Gleason score and 25 % or greater increase in cancer core involvement

	Univariate and multivariate analyses to predict upgrade of Gleason score			Univariate and multivariate analyses to predict 25 % or greater increase in cancer core involvement
	Univariate	Multivariate		Univariate	Multivariate
	p value	p value	Relative risk (95 % CI)	p value	p value	Relative risk (95 % CI)
Age (≦61 vs. 61<)	0.2	0.1	0.728 (0.102–7.468)	0.1	0.6	0.012 (0.005–0.467)
Clinical stage (T1c vs. T2a)	0.8	0.2	0.614 (0.208–6.080)	0.8	0.1	0.678 (0.044–2.880)
PSA (≦5 ng/ml vs. 5 ng/ml<)	0.1	0.5	0.388 (0.102–2.308)	0.2	0.2	0.352 (0.012–4.432)
PSA velocity (≦0 ng/ml/year vs. 0 ng/ml/year<)	0.1	0.9	0.038 (0.002–1.032)	0.020*	0.1	0.682 (0.042–3.782)
Gleason score at first diagnosis (3 + 3 = 6 vs. 3 + 4 = 7)	0.016*	0.06	0.858 (0.030–11.260)	0.9	0.1	0.718 (0.040–4.864)
% core of cancer (%) (at entry) (≦10% vs. 10%<)	0.005*	0.05	0.914 (0.242–8.360)	0.2	0.3	0.241 (0.008–2.026)
Risk group at start of AS (low vs. intermediate)	0.4	0.3	0.480 (0.204–2.218)	0.4	0.7	0.052 (0.088–1.026)
Patient with 5ARI (use vs. no use)	0.07	0.4	0.420 (0.204–2.074)	0.2	0.3	0.232 (0.341–2.554)
Change of diameter of biopsy proven index lesion (25 % or greater increase versus <25 % increase)
Major axis	0.047*	0.3	0.478 (0.232–1.878)	0.012*	0.022*	6.672 (1.097–40.508)
Minor axis	0.1	0.2	0.602 (0.302–8.018)	0.029*	0.1	0.728 (0.142–3.612)
Change in Doppler grade of index lesion (upgrade vs. stable or downgrade)	0.002*	0.037*	7.148 (1.809–63.145)	0.011*	0.039*	4.091 (1.673–24.878)

^*Statistically significant

In univariate analyses for predicting progression in over 25 % increase in cancer core involvement, PSA velocity (p = 0.020), increase (≥25 %) in major axis diameter (p = 0.012) and minor axis diameter (p = 0.029),and upgrade of Doppler grade (p = 0.011) were significant predictors of progression. In multivariate analyses, increase (≥25 %) in major axis diameter (hazard ratio, 6.672; 95 % CI, 1.097–40.508; p = 0.022) and the upgrade of Doppler grade (hazard ratio, 4.091; 95 % CI, 1.673–24.878; p = 0.039) were significant predictors of progression (Table 2).

Discussion

We first demonstrated that modern imaging modalities such as multi-parametric TRUS could have a role as monitoring tools to demonstrate possible progression of the visible baseline cancer by “per-lesion-based active surveillance” instead of whole-gland active surveillance. In our 11-year experience of active surveillance, we analyzed data for men who had undergone TRUS monitoring of the identical visible lesion, at least three times, which could be compared with the initial TRUS with three-dimensional schematic mapping of the lesion in the prostate, with a median follow-up of 4.4 years.

The surveillance prostate biopsy has been regarded as useful evaluation tool for AS. However, the sampling error of surveillance biopsy is well documented in many AS studies [18–20]. Berglund et al. [18] found upgrading in 27 % of the patients undergoing immediate restaging biopsy. Eggener et al. [19] found a 30 % rate of upgrading on restaging biopsy before initiation of surveillance-based treatment. van den Bergh et al. also reported that a 22 % of upgrading with re-biopsy at 1 year in a large cohort of more than 500 men undergoing active surveillance. On the other hand, the prostate cancers detected by image-guided targeted biopsies are of higher grade and large volume and are therefore likely to be clinically significant [7, 9, 15, 16]. In this study, we focused on the monitoring and analyzing of biopsy-proven image-visible lesions with special reference to their specific image-related variables (such as diameter of hypo-echoic lesion or Doppler grade of blood flow) in relation to the surveillance biopsy-proven progression in either cancer grade or cancer volume (such as Gleason score or cancer core involvement). Importantly, the sampling error in our study could be further minimized in comparison with the general systematic random biopsy strategy, because our surveillance biopsy was precisely sampled from the center of the TRUS-visible lesion by the longitudinally monitored image-based targeted biopsy. The reliable precision of the documented image-based targeted sampling is the essential technical key to support our new findings to demonstrate the significant relation between the image-related variables and image-targeted surveillance biopsy-proven progression.

In the present study, the upgrade of Doppler grade of biopsy-proven index lesion (hazard ratio, 7.148; 95 % CI, 1.809–63.145; p = 0.037) was significant predictor of progression with Gleason grade in multivariate analyses. Twenty-five percent or greater increase in major diameter of biopsy-proven index lesion (hazard ratio, 6.672; 95% CI, 1.097–40.508; p = 0.022), and the upgrade of Doppler grade of biopsy-proven index lesion (hazard ratio, 4.091; 95 % CI, 1.673–24.878; p = 0.039) were significant predictors of progression with upgrade of over 25 % in multivariate analyses. In the relationship between Gleason score and blood flow, previous studies reported positive correlation. Mitterberger et al. [21] reported that the Gleason score of all cancers detected on contrast-enhanced color Doppler targeted biopsy was higher (mean 6.8, n = 180) rather than the Gleason score of all cancers detected on systematic biopsy (mean 5.4, n = 166) (p < 0.003). Zhao et al. also reported that the blood flow grading correlated with Gleason score significantly (n = 43, r = 0.61, p < 0.001) [22]_{. In the relationship between the size of TRUS-visible} lesion and % core of cancer, Ukimura et al. reported that the positive correlation between the size of TRUS-visible lesion and cancer-involved core length or percent of core with cancer was found in 93 patients, a total of 681 biopsy cores [9]. These previous reports suggested the significant correlation between the image-related variables of lesion and aggressiveness of the cancer in either its grade or volume at the single time point at biopsy. In contrast, our new finding of the present study demonstrated that the longitudinal change of quantitative image-related variables (such as diameter of hypo-echoic lesion or Doppler grade of blood flow) could be a clinical predictor for the disease progression in their longitudinal analysis over different time points. The longitudinal documentation of the quantitative image-related variables in the identical biopsy-proven lesion is promising on AS program, and also allows reliable biopsy sampling by per-lesion-based surveillance.

However, the present study has several limitations. First, there is a limitation of visibility by multi-parametric TRUS in patients who had a very low-volume cancer, suitable for AS. When comparing TRUS-visible with image-invisible index lesions, the cancer-involved core lengths were 6.1 versus 1.5 mm (p < 0.001), respectively [9]. Multi-parametric TRUS-visible cancer at diagnostic biopsy likely has a higher volume than image-invisible cancer. Second, it should be noted that the entire measurable area of image-visible lesions does not perfectly correspond with the pathological cancer lesion. This means that underestimation and overestimation of cancer volume often happen when using imaging [23, 24]. As there is consensus that the threshold volume of clinically significant cancer was 0.5 ml [23], the multi-parametric TRUS has limitation to monitor the clinically significant cancer, which might be missed or underestimated by imaging. Clearly, a targeted biopsy-based estimation of cancer volume/aggressiveness is necessary in addition to image-based monitoring. Third, the operator dependency of TRUS imaging: the reproducibility of the TRUS findings in the present study in the hands of a less experienced operator is not known. As suggested in the literature, the differences in prostate cancer detection by TRUS among various operators are probably related to differences in expertise and/or technique [25]. The increase in cancer core involvement might be simply caused by sampling from the different parts of a lesion; as such, the progression in Gleason score could be a more reliable marker. On the other hand, based on the increased evidence of the utility of multi-parametric MRI, future use of multi-parametric MRI is promising for the possible further visibility of cancers to enhance per-lesion-based surveillance [24, 26–28]. Margel et al. [10] reported that the chance of reclassification on repeat biopsy was extremely low at 3.5 % when no cancerous lesion was identified on multi-parametric MRI. In addition, they reported the positive and negative predictive values of multi-parametric MRI to predict reclassification (83, 95 % CI 73–93 and 81, 95 % CI 71–91, respectively) in an unselected group of patients before AS [10]. Fradet et al. [29] reported that in the context of AS, a lesion suggesting cancer on MRI may confer a threefold increased risk of overall cancer progression. Cabrera et al. [30] reported that MRI without DWI does not appear to have prognostic value for AS. Finally, limitation of our study included the lack of comparison of biopsy-proven index cancer with final pathology of whole-gland prostate. Without pathological analysis of whole-gland prostate, it is difficult to exclude the possibility that a clinically important cancer has been missed.

In conclusion, longitudinal documentation of the TRUSvisible biopsy-proven cancer provides a new opportunity to perform per-lesion-based active surveillance for prostate cancer. Twenty-five percent or greater increase in major axis diameter and upgrade of Doppler grade of biopsy-proven lesion were significant predictors of biopsy-proven progression in the patients on AS. As long as the biopsy-proven cancer lesions were visible with modern multi-parametric TRUS imaging, per-lesion-based surveillance of the lesion could be reliably achieved with surveillance TRUS-guided targeted biopsy in combination with the longitudinally documented 3D mapping of the identical visible lesion.