SEARCH

SEARCH BY CITATION

Keywords:

  • prostate cancer;
  • MRI;
  • PSA;
  • biochemical recurrence;
  • clinical stage;
  • radical prostatectomy

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. Conflict of Interest
  9. References

Objective

  • To evaluate the suitability of preoperative multiparametric magnetic resonance imaging (MRI) positivity as a predictor of biochemical recurrence after radical prostatectomy (RP).

Patients and Methods

  • We reviewed the clinical records of patients who underwent either standard RP or laparoscopic RP between January 2005 and December 2009 at our institution.
  • Patients who received radiotherapy or androgen deprivation therapy before surgery were excluded. A total of 314 patients met the study inclusion criteria.
  • Cox proportional hazard regression models were used for analyses.
  • In accordance with the criteria in the established guidelines, a radiologist scored the probability of the presence of prostate cancer using a five-point scale of diagnostic confidence level. The highest confidence level of any pulse sequence was considered as the evaluation result.

Results

  • MRI positivity was significantly associated with a high clinical stage (cT ≥ 2; P = 0.039), a high positive biopsy core rate (≥0.2; P < 0.001), a high biopsy Gleason score ([GS] ≥8; P < 0.001) and a high pathological GS (≥8; P = 0.005).
  • Univariate analysis and multivariate analysis showed that MRI positivity was a prognostic indicator in the analysis that included only preoperative variables and also in the analysis including preoperative and pathological variables.

Conclusion

  • Multiparametric MRI positivity can independently predict biochemical recurrence after RP.

Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. Conflict of Interest
  9. References

Prostate cancer (PCa) is the most common cancer in men [1]. Life expectancy in patients with a high Gleason score (GS) PCa, managed with non-curative intent, is substantially reduced [2] so the correct stratification of patients into low- or high-risk groups is of utmost importance [3-5]. As biopsy-determined GS sometimes provides an inaccurate reflection of the final GS as determined in radical prostatectomy (RP) specimens (undergrading in 34–38% of cases) [6-8], it is also critical to obtain additional diagnostic information.

Multiparametric MRI has been increasingly incorporated into oncological imaging [9]. In several studies on PCa, MRI methods such as diffusion-weighted imaging (DWI) have been used as a diagnostic tool, with promising results [10-15]. In our institution, we previously evaluated and reported on the diagnostic ability of multiparametric MRI for PCa [16]; however, although the diagnostic ability of multiparametric MRI is recognized as very important and has been well discussed, only a few studies with a small number of cases have reported on the prognostic value of the multiparametric MRI positivity as an imaging biomarker. In the present study, we analysed the preoperative and peri-operative characteristics of patients who had undergone RP to evaluate the prognostic value of multiparametric MRI positivity as a disease-specific prognostic factor after RP.

Patients and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. Conflict of Interest
  9. References

We reviewed the medical records of patients who underwent laparoscopic or open RP between January 2005 and December 2009 at our institution. Clinical stage was assessed by DRE, abdominal and pelvic CT and bone scan, in accordance with the 2002 TNM classification system. All patients underwent MRI before surgery. An interval of at least 6 weeks was allowed between the biopsy and MRI to prevent the possible interference of post-biopsy haemorrhage. Patients who received radiotherapy or androgen deprivation therapy before surgery were excluded from the study, leaving a total study population of 314 patients, with a mean (range) age of 64.3 (43–76) years. The mean (range) follow-up period was 61.1 (24.2–95.6) months. PCa was histologically confirmed before surgery by TRUS-guided, non-targeted needle biopsy (six to 20 biopsies, median: 10). The vast majority of our patients had their initial biopsies, where the extended biopsy schema was applied. Immediately after RP, specimens were fixed in 10% neutral buffered formalin. The entire specimen was serially cut at 4-mm intervals. The tumour border was outlined on the coverslip of the slide using a marking pen. A genitourinary pathologist (S.M.), with > 15 years of experience and blinded to the MRI results, interpreted the pathological outcome. Tumours were graded according to the 2005 International Society of Urological Pathology Modified Gleason Grading System. Histopathological analysis showed that 201 men (64.0%) had disease confined to the prostate (pT2), and 113 (36.0%) had extraprostatic disease (pT3). Positive surgical margins (PSMs) were found in 77 patients (24.5%); the pathological GS range was 5–9 (Table 1). Biochemical recurrence occurred in 36 patients overall. No patient was found to have local recurrence or metastatic lesions after biochemical recurrence.

Table 1. Patients' peri-operative and pathological characteristics
CharacteristicOverallMRI-negativeMRI-positiveP*
  1. *Student t-test and chi-squared test.

No. of patients31483231 
Mean (sd) age, years64.3 (6.2)63.7 (6.4)64.6 (6.0)0.258
Mean (sd) BMI, kg/m223.7 (2.6)23.7 (2.7)23.7 (2.5)0.963
Mean (sd) survival, months52.2 (21.3)49.5 (17.9)53.2 (22.3) 
Mean (sd) Watch, months56.8 (18.5)51.1 (16.8)58.8 (18.7) 
Mean (sd) preoperative PSA, ng/mL7.7 (3.8)7.0 (3.6)8.0 (3.8)0.052
Preoperative PSA, n    
<10 ng/mL25272180 
>10, <20 ng/mL56947 
>20 ng/mL624 
Biopsy GS   <0.001
≤61495693 
713723114 
≥828424 
Clinical T stage   0.039
cT1c28580205 
cT229326 
Biopsy positive core rate   <0.001
<0.21315279 
≥0.218331152 
Surgical margin status   0.687
Negative23764173 
Positive771958 
Pathological GS   0.005
≤6461927 
722860168 
≥840436 

In a previous study by researchers from our institution, patients underwent preoperative MRI using a 1.5-T MR scanner (Signa Excite XI 1.5T, 8-channel torso-array coil) between 2005 and 2009. T2-weighted fast spin-echo imaging (5000/87.9, 4 NEX, 3 min 45 s), DWI (single-shot diffusion echo planar imaging, b = 0, 1000 s/mm2, 3600/72.6, 1 min and 55 s), and Gadolinium-dynamic MRI (fat-suppressed fast Spoiled Gradient Recalled Echo (SPGR), 130/2.0/90o, 22 s, 40, 80, and 180 s after i.v.) were performed. For DWI, apparent diffusion coefficient (ADC) maps and exponential ADC maps were constructed. One experienced radiologist (A.T.), with >20 years of experience at reading prostate magnetic resonance images, reviewed and analysed the images. The radiologist knew the patients had clinically diagnosed PCa but was blinded to patient-specific clinical data and biopsy findings.

Using the following criteria in accordance with the established guidelines [17], the radiologist scored the probability of the presence of PCa (for both peripheral and transition zones) based on a five-point diagnostic confidence level scale (5: definitely positive; 4: probably positive; 3: possibly positive; 2: probably negative; 1: definitely negative). The highest confidence level of any pulse sequence was used as the evaluation result. Scores of 5, 4, and 3 were defined as positive findings, while scores of 1 and 2 were considered to be negative findings (Fig. 1).

figure

Figure 1. Five-point diagnostic confidence levels of PCa.

Download figure to PowerPoint

Diagnostic Criteria

Any lesion in the peripheral zone showing a decrease in the ADC was considered malignant (Fig. 1, part a1: level 5). If the lesion showed a warm colour (Fig. 1, part a2: red colour) on the eADC [eADC = 1/exp(b*ADC)] map, it was assigned level 5. If the lesion was located in the transition zone but the colour was apparently cold or warm (Fig. 1, part b2), or was small (Fig. 1, part c2), it was assigned level 4 or 3. If lesions <5 mm in size were located in the transition zone and the colour was moderately cold or warm, they were not considered and were assigned level 2 or 1 (Fig. 1, parts d2, e2). Low signal intensity in the peripheral zone, accompanied by disruption of the duct structure, was considered malignant (Fig. 1, parts f, g: levels 5 or 4). In the transition zone, an irregular low signal intensity area without a capsule, or a signal intensity obliterating normal structures, such as the surgical capsule or the urethra, were considered malignant (level 5). Moderately low signals in the peripheral zone or transition zone were considered possibly malignant (Fig. 1, part h: level 3). If low signal intensities on T2-weighted and high-density T1-weighted images were noted, the lesion was considered to be benign (level 2). Symmetrical wedge-shaped low signal intensity with a maintained duct structure was considered non-cancerous (Fig. 1, part i: level 2). Lesions showing enhancements in the early phase and a washout in the delayed phase were considered positive (Fig. 1, parts k1 and k2: level 5). Lesions showing early enhancement and a weak washout in the delayed phase were suspected to be malignant (Fig. 1, parts l1 and l2: level 4). Lesions showing early and prolonged enhancement were suspected of being malignant (Fig. 1, parts m1 and m2: level 3). Lesions that did not show early enhancement but showed a gradual enhancement in the delayed phase were not considered to be positive (Fig. 1, parts n1 and n2: level 2). Figure 1, parts j, o1 and o2 were considered to be BPH (Fig. 1, parts j, and o1: level 1).

Serum PSA levels were measured at least every 3 months after surgery. Biochemical recurrence was defined as an increase or persistence of serum PSA levels > 0.2 ng/mL after surgery. Cox proportional hazard regression models were used for univariate and multivariate analyses to test the relationships between biochemical recurrence and pathological factors. For statistical analyses, the value that best discriminated between ‘good’ and ‘poor’ biochemical recurrence-free survival was found by testing all possible thresholds within the central 80% of the distribution of values. P values for the clinically relevant thresholds were corrected for multiple testing. To obtain a multivariate model with maximum precision for the important variables, a stepwise selection procedure was used. Using the statistically significant variables in the multivariate Cox regression analysis, the odds ratio (OR) of biochemical recurrence was estimated. Biochemical recurrence-free probabilities for each group were estimated using the Kaplan–Meier method. Biochemical recurrence was compared between groups using a log-rank test. In all tests, differences were considered significant at P < 0.05. The analyses were performed using spss, version 20.0 (SPSS Inc, Chicago, IL, USA). Data were collected according to a Keio University School of Medicine Institutional Review Board protocol (Approval no. 2012-272).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. Conflict of Interest
  9. References

Our results showed that MRI positivity was significantly associated with a positive DRE (P = 0.039), a high positive core rate (P < 0.001), a high biopsy GS (P < 0.001) and a high pathological GS (P = 0.005; Table 1).

To examine the relationship between biochemical recurrence and preoperative variables, we first performed univariate and multivariate analyses of factors predicting PSA recurrence for preoperative variables only. In the univariate Cox proportional hazard model, clinical T stage (assessed by DRE), MRI positivity, biopsy GS ≥ 8 and positive core rate > 0.2 were significantly associated with biochemical recurrence. In the multivariate Cox proportional hazard model using stepwise inclusion of these variables, positive core rate > 0.2, biopsy GS ≥ 8 and MRI positivity were significantly associated with biochemical recurrence (Table 2). In the present study, preoperative PSA was not a significant variable associated with biochemical recurrence.

Table 2. Statistical analysis of factors predicting biochemical recurrence
VariableUnivariate analysisMultivariate analysis
POR (95% CI)P
Clinical T stage0.039  
cT1c   
cT2   
MRI0.006 0.031
Negative 1 
Positive 5.05(1.16–22.05) 
Biopsy positive cores0.004 0.012
≤0.2 1 
>0.2 3.19(1.29–7.88) 
Biopsy GS<0.001 0.001
≤7 1 
≥8 4.85(1.89–12.46) 

An additional analysis of preoperative and postoperative factors (including histopathological findings) was performed to identify factors predicting postoperative biochemical recurrence. In the univariate Cox proportional hazard model, clinical T stage, MRI positivity, biopsy GS ≥ 8, pathological GS ≥ 8, positive core rate >0.2 and PSMs were significantly associated with biochemical recurrence. In the multivariate Cox proportional hazard model using stepwise inclusion of these variables, MRI positivity, pathological GS ≥ 8 and PSMs were independent variables associated with biochemical recurrence (Table 3).

Table 3. Statistical analysis of factors predicting biochemical recurrence (including preoperative and pathological variables)
VariableUnivariate analysisMultivariate analysis
POR (95% CI)P
Clinical T stage0.039  
cT1c   
cT2   
MRI0.006 0.016
Negative 1 
Positive 6.19 (1.41–27.08) 
Biopsy GS<0.001  
≤7   
≥8   
Biopsy positive cores0.004  
≤0.2   
>0.2   
Pathological GS<0.001 0.020
≤7 1 
≥8 3.38 (1.44–7.93) 
Pathological T stage<0.001  
pT2   
pT3   
Surgical margin status<0.001 0.001
Negative 1 
Positive 3.89 (1.83–8.26) 

Using the three preoperative variables found to be significant in multivariate Cox regression analysis (biopsy GS, biopsy core rate and positivity on MRI), we developed a prognostic factor-based model for risk stratification (Fig. 2). The OR of biochemical recurrence was calculated as exp [(1.16 × positive core rate) + (1.61 ×MRI positivity) + (1.58 × biopsy GS)]. In this equation, a positive core rate ≤0.2 and positive core rate >0.2 equaled 0 and 1, respectively; negativity and positivity on MRI equaled 0 and 1, respectively; and biopsy GS ≤ 7 and ≥8 equaled 0 and 1, respectively. Based on the OR of biochemical recurrence, patients were divided into three risk groups: low (relative risk [RR] 1.00–3.19, 78 men), intermediate (OR 4.85–5.05, 84 men), and high (OR 15.47–78.13, 152 men). The low-risk group consisted of patients with zero or one risk factor (positive core rate >0.2). The intermediate-risk group consisted of patients with one risk factor between biopsy GS ≥ 8 and MRI positivity. All others patients were included in the high-risk group. The 5-year biochemical recurrence-free rates were 100% in the low-risk group, 91% in the intermediate-risk group, and 84% in the high-risk group. The differences across groups were significant (low-risk vs intermediate risk: P = 0.011, intermediate risk vs high-risk: P = 0.029, low-risk vs high-risk: P < 0.001)

figure

Figure 2. Kaplan–Meier curve according to risk category (including preoperative factors).

Download figure to PowerPoint

Using the three preoperative variables significant in multivariate Cox regression analysis (pathological GS, surgical margin and MRI positivity), we also developed a prognostic factor-based model for risk stratification among risk factors including pathological variables (Fig. 3). The OR of biochemical recurrence was calculated as exp [(1.36 × PSM) + (1.82 ×MRI positivity) + (1.22 × pathological GS)]. In this equation, negativity and positivity on surgical margin equaled 0 and 1 respectively; negativity and positivity on MRI equaled 0 and 1, respectively; and pathological GS ≤ 7 and ≥8 equaled 0 and 1, respectively. Based on the OR of biochemical recurrence, patients were divided into three risk groups: low (RR 1.00–3.89, 80 men), intermediate (OR 6.19–13.15, 148 men), and high (OR 20.92–81.39, 86 men). The low-risk group consisted of patients with zero or one risk factor (PSM, pathological GS ≥ 8). The intermediate-risk group consisted of patients with one risk factor (MRI positivity) or two risk factors (PSM, pathological GS ≥ 8). All others patients were included in the high-risk group. The 5-year biochemical recurrence-free rates were 99% in the low-risk group, 92% in the intermediate-risk group, and 72% in the high-risk group. The differences across groups were significant (low-risk vs intermediate risk: P = 0.025, intermediate risk vs high-risk: P = 0.002, low-risk vs high-risk: P < 0.001).

figure

Figure 3. Kaplan–Meier curve according to risk category (including pathological factors).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. Conflict of Interest
  9. References

In the American Joint Committee on Cancer guidelines 6th edition, clinical stage was determined by DRE, while in the 7th edition, both DRE and ‘visible imaging’ were considered to determine clinical stage. Clinical stage should adequately reflect the influence of prognostic variables. Despite the introduction of MRI into clinical staging, few analyses of the direct relationship between MRI positivity and disease-specific prognosis have been conducted.

In the present study, MRI positivity was significantly associated with DRE positivity, high positive core rate, high biopsy GS and high pathological GS (Table 1). As shown in Table 1, 70% of DRE-negative (cT1c by DRE) patients were MRI-positive (cT2 or above), while 10% of DRE-positive patients were MRI-negative. Although DRE and MRI findings were found to have a strong significant association (P= 0.039), clinical stage varied greatly depending on whether or not MRI positivity was included.

Previous retrospective studies have reported that within the prostate, DWI can improve the pretreatment prediction of true GS [10, 18]. The significant reduction of ADC values in PCa has been well documented [16, 19-21]. The predominant contribution of DWI signals and the decrease in ADC values in malignant tissue has been attributed to histopathological characteristics, including hypercellularity, enlargement of nuclei, hyperchromatisms and angulation of the nuclear contour, all of which reduce the diffusional displacement of water molecules [22].

It has been reported that there is a significant relationship between tumour volume and biochemical recurrence [23-25]. The percentage of positive cores has been reported to be a strong predictor of tumour volume [26, 27], and consistent with these studies, positive core rate was found to be significantly correlated with MRI positivity in our study (Table 1). The results of our study are in accordance with a previous study that concluded that the detection of PCa by T2-weighted imaging is significantly dependent on the lesion GS and tumour size [27]. Further investigations are needed to analyse the anatomical correlation between the sites of MRI positivity and tumour location.

Although the diagnostic ability of MRI is very important and well discussed, only a few reports with a small number of cases have discussed the prognostic value of MRI positivity as an imaging biomarker. In the present study, we attempted to evaluate the value of multiparametric MRI, not only as a diagnostic tool of PCa, but also as an independent biomarker of biochemical recurrence. Multiparametric MRI positivity was found to be an independent prognostic factor of biochemical recurrence after RP, both among preoperative variables and among preoperative and pathological variables. For the clinician, simple and available factors which can directly predict biochemical recurrence are very important, both for preoperative and follow-up strategy.

In terms of the clinical impact of the present findings, MRI positivity is an independent, brief and non-invasive biomarker which can predict therapeutic effects preoperatively. The predictive ability is as high as biopsy GS and positive core rate, and both of these are invasive biomarkers. In other recent reports, MRI is taken as a non-invasive, alternative means of biopsy GS [28, 29]. The present study provides supportive evidence for this logic.

Among postoperative variables, MRI positivity was an independent factor that could predict biochemical recurrence, and this means that MRI positivity may have the potential to be an effective index for determination of a therapeutic strategy, as well as a follow-up strategy. This can be of benefit for clinicians as a strong and easy-to-use predictive factor. Further study should be conducted to determine a concrete strategy that is suitable for each risk group.

The present study has several limitations. Its retrospective nature may have resulted in some evaluation bias as patients with adverse features on MRI may have been offered alternative therapies. Prospective joint trials with other institutions are needed to evaluate the validation of multiparametric MRI as an imaging biomarker which predicts biochemical recurrence after RP. In addition, although the radiologist was blinded to patient-specific clinical data and biopsy findings, he was aware of the presence of clinically determined PCa.

In conclusion, this study demonstrated that multiparametric MRI positivity is an independent prognostic factor of biochemical recurrence after RP. In addition to its existing value in diagnostic imaging, MRI positivity itself as a non-invasive imaging biomarker may contribute to the assessment of PCa prognosis.

Acknowledgement

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. Conflict of Interest
  9. References

We thank all participants in the study. The corresponding author (T.K.) had full access to all data and had final responsibility for the decision to submit the manuscript for publication.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Acknowledgement
  8. Conflict of Interest
  9. References
Abbreviations
PCa

prostate cancer

GS

Gleason score

RP

radical prostatectomy

mp

multiparametric

DWI

diffusion-weighted imaging

PSM

positive surgical margin

ADC

apparent diffusion coefficient

OR

odds ratio

RR

relative risk