Abstract
Purpose
Prostate-specific membrane antigen (PSMA) positron emission tomography and multiparametric magnetic resonance imaging (mpMRI) are used to contour the dominant intraprostatic lesion (DIL) when incorporating a microboost. However, it is unclear how well PSMA and mpMRI findings correlate with the true pathologic DIL. So, we performed an analysis between preoperative imaging and the final surgical specimen to identify predictors of discordance.
Methods and Materials
This was a single-center retrospective analysis of patients with prostate cancer who underwent PSMA imaging before prostatectomy between January 2022 and December 2023. PSMA lesions were independently contoured and measured by a reviewer blinded to the final pathology. mpMRI reports were reviewed for lesion number, location, and size. Lesion size and hemigland location on imaging were compared to the final pathology. Associations between concordance or discordance and patients’ demographics and clinicopathological characteristics were examined.
Results
Of the 48 patients included, mean (SD) PSA was 13.9 (±12.9), 40% had grade group 4/5 disease, and 58% were staged as pT3 or higher. The pathologic DIL matched the hemigland of the largest PSMA/mpMRI lesion in 94/90% of cases, respectively. The median maximum lesion diameter size discrepancy between PSMA and mpMRI and final pathology was 3.4 mm (IQR, 16.3-24.8) and 4 mm (IQR, 3-9), respectively. PSMA and mpMRI underestimated lesion size in 59% and 64% of cases, respectively. On PSMA, higher prostate-specific antigen (PSA) and lower biopsy grade groups were associated with greater imaging/pathology size discrepancy on both univariate and multivariable analysis after adjusting for maximum standardized uptake volume (SUVmax); on mpMRI, higher PSA was associated with greater imaging/pathology size discrepancy on both univariate and multivariable analysis after adjusting for prostate imaging reporting and data system (PI-RADS) score (P < .05).
Conclusion
Standard treatment margins may either overestimate or underestimate the extent of the true gross disease, depending on individual tumor characteristics. Further research is needed to develop a standardized framework for using PSMA and mpMRI to guide microboosting.
Introduction
A meta-analysis of 18 randomized prostate cancer trials suggests that local failure after definitive radiation occurs in approximately 10% of cases.1 In other studies that have used more sensitive methods of detecting a local recurrence, such as with prostate-specific membrane antigen (PSMA) imaging, intraprostatic only recurrences may be as high as 35%.2 In cases where there is an intraprostatic failure, the most likely site of recurrence tends to be within the dominant intraprostatic lesion (DIL).3 To reduce in-field failures, the use of a simultaneous integrated boost to the DIL, commonly referred to as a microboost, was shown to improve disease-free survival in a prospective randomized controlled trial.4 However, there have been significant challenges to adopting the microboost approach in clinical practice such as accurately defining the DIL on imaging and identifying the appropriate clinical target volume (CTV) margin.5 Given a study where DIL boost contour inaccuracies predicted an increased risk of biochemical failures related to undercoverage of the target to full boost dose,6 there is an unmet need to standardize contours and CTV expansions to achieve consistent outcomes with microboosting.
Multiparametric magnetic resonance imaging (MRI) (mpMRI) is the preferred modality to use for fusion biopsies and is commonly used to define the DIL for microboosting. However, there are multiple limitations to mpMRI such as variability in imaging quality and reliability of radiographic interpretation.7 Alternatively, PSMA positron emission tomography (PET) imaging has a growing role in the staging of newly diagnosed prostate cancer.8 An inherent advantage to PET imaging is a direct visualization of PSMA-avidity, which may facilitate the contouring of a DIL for the microboost. To this end, we sought to evaluate concordance rates between the DIL on PSMA PET, mpMRI, and surgical pathology and to identify predictors of discordance between these 3 modalities with the goal of refining target delineation for microboosting.
Methods and Materials
Study design and patients
This was a single-center retrospective study approved by the Cedars-Sinai Medical Center Institutional Review Board. We included patients with prostate cancer who underwent radical prostatectomy between January 2022 and December 2023 with an fluorine-18-labeled DCFPyL (18F-DCFPyL) PSMA PET within 4 months of surgery. We excluded patients with missing prostatectomy pathology report, missing PSMA PET report, carcinoma in situ disease, those who received androgen deprivation therapy before surgery, no DIL identified on final pathology, or with an inaccessible PSMA PET for direct measurement (Fig. 1). Patients were not required to undergo mpMRI before surgery.
Figure 1.
Patient selection flowchart.
Abbreviations: ADT = androgen deprivation therapy; CT = computed tomography; PET = positron emission tomography; PSMA = prostate-specific membrane antigen.
PSMA PET/computed tomography (CT) and mpMRI acquisition
PSMA PET/CT images were acquired after intravenous injection of 18F-DCFPyL PSMA using a Siemens Biograph PET-CT Fusion Scanner from the base of the brain to the mid-thigh. The CT scan was used for attenuation correction and anatomic correlation only. A Siemens 3 T magnet (n = 22) and Siemens 1.5 T magnet (n = 1) were used for mpMRI. For the remaining patients, the magnet strength was not specified (18 patients) or they did not receive an MRI (7 patients).
Image analysis
A blinded reviewer independently contoured all visible prostate cancer lesions on PSMA PET scans, determined which lesion was the largest, and recorded the maximum dimension of all lesions blinded to the final largest dimension recorded on the pathology report. A second reviewer also reviewed all PSMA PET scans and confirmed the largest diameter measurement. In cases of uncertainty in measurement, the 2 reviewers evaluated the imaging results until consensus was reached. The size and hemigland location of the DIL were abstracted from the final prostatectomy pathology report. The SUV max of the largest PSMA PET lesion in the same hemigland of the DIL was also recorded.
Tumor hemigland location on PSMA PET and final pathology were considered to match if they were in the same hemigland (left/right). Partial matches (ie, the PET only noted a lesion on the left side of the prostate but the final tumor was located in the left prostate but also crossed over to the right side of the prostate) were considered a match for univariate analysis. In addition to a hemigland match, we also verified whether the lesion matched in terms of being within either the base, mid, or apex of that prostate hemigland. This was done by an author with over 15 years of experience in treating prostate cancers. For comparisons of the size of the maximum lesion diameter on imaging, we used the largest size in the same hemigland as the DIL on prostatectomy even if there was a larger lesion in the contralateral hemigland on imaging.
For the mpMRI lesion analysis, the following metrics were abstracted from the imaging reports: number of lesions, the greatest diameter of the lesions, prostate volume, PSA density, and PI-RADS score. The MRI lesion with the largest maximum dimension that was in the same hemigland as the DIL on the final pathology report was used for analysis.
Statistical analysis
Data are presented as the number of patients (%, percentage) for categorical variables, and mean (± SD) and median (IQR) for continuous variables.
Univariate and multivariable analyses of absolute differences between imaging and pathologic diameter on PET and on MRI were examined using a generalized additive model for location, scale, and shape with a skew t type 1 distribution and t family distribution with an identity link function, respectively.9 The goodness of fit of each model was examined using residuals such that the most adequate response distribution is chosen.10 Multivariable analysis was performed using a stepwise variable selection procedure based on the generalized Akaike information criterion. Clinicopathologic characteristics considered for inclusion included PSA, PSA density, Max biopsy Gleason score, PI-RADS score of the biggest MRI lesion in prostatectomy hemigland DIL, and SUV max of lesion on PET in the same hemigland as prostatectomy, which were chosen a priori.
Analyses were performed using R software (v4.4.3) with 2-sided tests at a significance level of 0.05.11 Significance was not adjusted for multiple tests.
Results
Baseline characteristics
In total, 48 patients (mean [SD] age, 65.85 [± 6.73]) met the study inclusion criteria (Fig. 1). Patient characteristics and clinicopathological variables are summarized in Table 1. The mean (SD) PSA at diagnosis was 13.9 (± 12.9) and 45% (22 of 48 patients) had grade group 4 or 5 disease on initial biopsy. On final surgical pathology, 40% (19 of 48 patients) had grade group 4 or 5 disease and 58% (28 of 48 patients) were staged as T3a or higher. MRI imaging was more commonly ordered before PSMA PET, as the median (IQR) time from MRI to surgery was 108 (66-176) days and the median (IQR) time from PSMA PET to surgery was 44 (21.5-69) days.
Table 1.
Patient characteristics
| Variable | N = 48 |
|---|---|
| Age (y) | |
| Mean (±SD) | 65.85 (± 6.73) |
| Median (IQR) | 67 (60.5-70.0) |
| Grade group on biopsy | |
| 1 | 1 (2) |
| 2 | 11 (23) |
| 3 | 13 (28) |
| 4 | 9 (19) |
| 5 | 13 (28) |
| Missing | 1 |
| Cribriform on biopsy | |
| No | 22 (54) |
| Yes | 19 (46) |
| Missing | 7 |
| Intraductal on biopsy | |
| No | 32 (78) |
| Yes | 9 (22) |
| Missing | 7 |
| PSA (ng/mL) | |
| Mean (± SD) | 13.9 (± 12.9) |
| Median (IQR) | 9.7 (6.4-15.4) |
| Missing | 3 |
| Pathologic stage | |
| 2 | 20 (42) |
| 3a | 16 (33) |
| 3b/3bN1 | 12 (25) |
| Percent prostate involved by tumor on final path | |
| Mean (± SD) | 24.27 (± 19.36) |
| Median (IQR) | 20 (10-35) |
| Cribiform pattern | |
| No | 9 (19) |
| Yes | 39 (81) |
| Intraductal component | |
| No | 25 (52) |
| Yes | 23 (48) |
| Cribriform pattern and intraductal component | |
| No | 26 (54) |
| Yes | 22 (46) |
| Lymphovascular invasion | |
| No | 38 (79) |
| Yes | 10 (21) |
| Perineural invasion | |
| No | 4 (8.33) |
| Yes | 44 (91.67) |
| Pathologic grade group | |
| 1 | 1 (2.08) |
| 2 | 15 (31.25) |
| 3 | 13 (27.08) |
| 4 | 7 (14.58) |
| 5 | 12 (25) |
| Time between MRI and PET (d) | |
| Median (IQR) | 55 (38-134) |
| Missing | 5 |
| Time between PET and prostatectomy (d) | |
| Median (IQR) | 43.5 (21.5-69) |
| Time between MRI and prostatectomy (d) | |
| Median (IQR) | 108 (66-176) |
| Missing | 5 |
Data are presented as number of patients (%), mean (±SD), or median (IQR).
Abbreviations: MRI = magnetic resonance imaging; PET = positron emission tomography; PSA = prostate-specific antigen.
PSMA PET findings
On final surgical pathology, in all but 1 case, the highest Gleason score was located within the DIL. We correlated the preoperative PSMA PET features with the final surgical pathology (Table 2). The pathologic DIL was in the same hemigland as the maximum diameter tumor on PSMA imaging in 94% (45 of 48) of cases when including partial matches and 73% (35 of 48) of cases when excluding partial matches. There were 2/45 (4%) cases in which there was a hemigland match but the lesion location within the hemigland was not a match. In both cases, the lesion on final pathology was in the base/mid-gland while the PSMA-positive area was in the apex. When considering concordant DIL between PSMA and mpMRI, the median (IQR) SUV max was 8.7 (6.3-19) and median (IQR) maximum diameter was 19.5 mm (16.3-24.8). The median (IQR) difference in size between pathologic and PSMA maximal dimension was 3.4 mm (1.7-6.1). The maximal diameter on PSMA PET was less than the pathological maximal dimension in 59% (26 of 44) of cases. This difference in diameter exceeded 5 mm in only 26% (12 of 47) of cases.
Table 2.
PSMA features and pathology correlations
| SUV max of lesion on PET in the same hemigland as prostatectomy | |
| Median (IQR) | 8.7 (6.3-19) |
| Missing | 1 |
| PSMA max diameter in the same hemigland as prostatectomy (mm) | |
| Median (IQR) | 19.5 (16.3-24.8) |
| Missing | 1 |
| Pathologic dominant nodule max dimension (mm) | |
| Median (IQR) | 20.5 (16-28) |
| Is pathologic dominant nodule hemigland in the same hemigland as the largest tumor on PSMA (partial as mismatch)? | |
| Yes | 35 (72.92) |
| No | 13 (27.08) |
| Is pathologic DIL hemigland in the same hemigland as the largest tumor on PSMA (partial as match)? | |
| Yes | 45 (93.75) |
| No | 3 (6.25) |
| Is pathologic DIL hemigland in the same hemigland as the largest tumor on MRI? | |
| Yes | 36 (90) |
| No | 4 (10) |
| Missing | 8 |
| Absolute difference in pathologic DIL max dimension and PSMA max diameter in the same hemigland as prostatectomy (mm) | |
| Mean (± SD) | 4.65 (± 4.49) |
| Median (IQR) | 3.4 (1.7-6.1) |
| Agreement (≤5) | 35 (74.47) |
| Disagreement (>5) | 12 (25.53) |
| Missing | 1 |
| Difference in pathologic DIL max dimension and PSMA max diameter in the same hemigland as prostatectomy (mm) | |
| PSMA max diameter < path DIL max dimension | 26 (59.09) |
| PSMA max diameter > path DIL max dimension | 18 (40.91) |
| Missing | 4 |
Data are presented as the number of patients (%), mean (± SD), or median (IQR).
Abbreviations: DIL = dominant intraprostatic lesion; MRI = magnetic resonance imaging; PET = positron emission tomography; PSMA = prostate-specific membrane antigen; SUV = standardized uptake value.
We identified 3 cases with discordant DIL locations between PSMA PET and surgical pathology. Two of these cases did not demonstrate PSMA-avidity within the prostate despite a dominant lesion on mpMRI. Of these 2 cases, 1 mpMRI lesion was located in the opposite hemigland from the DIL on surgical pathology. By contrast, the other case had an MRI lesion that corresponded to the true DIL location, which could have helped guide appropriate treatment despite the negative PET. The third patient had no MRI or biopsy information available.
Univariate and multivariable analysis of the absolute difference between imaging and pathologic diameter between PSMA DIL and pathological DIL was performed when considering cliniopathologic factors that would be available at the time of diagnosis (Table 3). On univariate analysis, higher levels of PSA and lower biopsy grade group were associated with a higher absolute difference between imaging and pathologic diameter on PSMA. On multivariable analysis, after adjusting for SUV max, higher PSA and lower biopsy grade group remained associated with higher absolute difference between imaging and pathologic diameter on PSMA.
Table 3.
Univariate and multivariable analyses of absolute difference between imaging and pathologic diameter on PET with clinicopathologic characteristics
| Clinicopathologic characteristic | Univariate |
Multivariable* |
||
|---|---|---|---|---|
| Estimate (95% CI) | P value | Estimate (95% CI) | P value | |
| PSA† | 0.071 (0.042, 0.099) | <.001 | 0.081 (0.062, 0.099) | <.001 |
| Grade group on biopsy | ||||
| 1/2/3 | 1.693 (0.521, 2.865) | .007 | 1.628 (0.873, 2.383) | <.001 |
| 4/5 | 0 (reference) | 0 (reference) | ||
| SUV max of lesion on PET in the same hemigland as prostatectomy† | 0.004 (−0.012, 0.019) | .635 | −0.011 (−0.032, 0.010) | .299 |
Abbreviation: CI = confidence interval; PET = positron emission tomography; PSA = prostate-specific antigen; SUV = standardized uptake value.
Model was fit using a skew t type 1 distribution with an identity link function including 44 observations.
Estimate is expressed as 1-unit increment.
Multiparametric MRI findings
We then evaluated mpMRI features and their correlation with final pathology (Table 4). The median (IQR) maximum diameter of the lesion in the same hemigland as the DIL on final pathology was 20 mm.12, 13, 14, 15, 16, 17, 18, 19, 20, 21 The MRI lesions with the maximum diameter matched the hemigland with the maximum SUV uptake on PET in 88% of cases. Among the cases where the hemigland was a match, all cases were also a match when localizing the lesion to the base/mid/apex of that hemigland. The mean (SD) PI-RADS score of the maximum diameter MRI lesion in the same hemigland of the DIL on prostatectomy was 4.6 (72). The median (IQR) absolute difference in the pathologic DIL maximum dimension and the MRI maximum dimension was 4 mm.3, 4, 5, 6, 7, 8, 9 In 64% of cases, the maximum diameter on the MRI was less than the maximum dimension of the DIL seen on final pathology.
Table 4.
MRI features and pathology correlations
| Variable | No (n = 3) | Yes (n = 45) | P value |
|---|---|---|---|
| Age (y) | |||
| Mean (± SD) | 63.33 (± 3.51) | 66.02 (± 6.88) | 0.51 |
| Grade group on biopsy | |||
| 1 | 0 (0) | 1 (2.22) | 0.07 |
| 2 | 2 (100) | 9 (20) | |
| 3 | 0 (0) | 13 (28.89) | |
| 4 | 0 (0) | 9 (20) | |
| 5 | 0 (0) | 13 (28.89) | |
| Cribriform on biopsy | |||
| No | 2 (100) | 20 (51.28) | 0.49 |
| Yes | 0 (0) | 19 (48.72) | |
| Intraductal on biopsy | |||
| No | 2 (100) | 30 (76.92) | 1.00 |
| Yes | 0 (0) | 9 (23.08) | |
| PSA | |||
| Median (IQR) | 18.03 (12.6-23.45) | 9.3 (6.08-15.36) | 0.22 |
| PET location | |||
| Cedars | 3 (100) | 34 (75.56) | 1.000 |
| Other | 0 (0) | 11 (24.44) | |
| SUV max of lesion on PET in the same hemigland as prostatectomy | |||
| Median (IQR) | 0 (0-3.3) | 8.9 (6.55-19.55) | 0.005 |
| PET max diameter in the same hemigland as prostatectomy (mm) | |||
| Median (IQR) | 0 (0-0) | 20.2 (16.7-24.8) | 0.02 |
| Days between MRI and PET | |||
| Median (IQR) | 146 (49-243) | 55 (38-125) | 0.44 |
| MRI max diameter in the same hemigland as prostatectomy (mm) | |||
| Median (IQR) | 17 (15-19) | 20 (15-24.5) | 0.56 |
| MRI lesion with the greatest diameter matches the hemigland of the greatest PET | |||
| No | 2 (100) | 3 (7.89) | 0.01 |
| Yes | 0 (0) | 35 (92.11) | |
| PI-RADS score of the biggest MRI lesion in prostatectomy hemigland DIL | |||
| Median (IQR) | 5 (5-5) | 5 (4-5) | 0.34 |
| PI-RADS score of the biggest lesion on MRI | |||
| Median (IQR) | 5 (5-5) | 5 (4-5) | 0.31 |
| Days between PET and prostatectomy | |||
| Mean (± SD) | 21.67 (± 18.45) | 48.84 (± 26.77) | 0.09 |
| Days between MRI and prostatectomy | |||
| Median (IQR) | 157.5 (66-249) | 108 (69-171) | 0.80 |
| Path stage on prostatectomy | |||
| 2 | 3 (100) | 17 (37.78) | 0.06 |
| 3a | 0 (0) | 16 (35.56) | |
| 3b/3bN1 | 0 (0) | 12 (26.67) | |
| Path DIL maximum dimension (mm) | |||
| Median (IQR) | 16 (1.5-16) | 21 (17-28) | 0.045 |
| Percent prostate involved by tumor | |||
| Median (IQR) | 5 (1-10) | 20 (10-35) | 0.02 |
| Final path cribriform | |||
| No | 2 (66.67) | 7 (15.56) | 0.09 |
| Yes | 1 (33.33) | 38 (84.44) | |
| Final path intraductal | |||
| No | 2 (66.67) | 23 (51.11) | 1.00 |
| Yes | 1 (33.33) | 22 (48.89) | |
| Final path cribriform + intraductal | |||
| No | 2 (66.67) | 24 (53.33) | 1.00 |
| Yes | 1 (33.33) | 21 (46.67) | |
| Final path lymphovascular invasion | |||
| No | 3 (100) | 35 (77.78) | 1.00 |
| Yes | 0 (0) | 10 (22.22) | |
| Final path perineural invasion | |||
| No | 2 (66.67) | 2 (4.44) | 0.02 |
| Yes | 1 (33.33) | 43 (95.56) | |
| Grade group final path | |||
| 1 | 1 (33.33) | 0 (0) | 0.02 |
| 2 | 2 (66.67) | 13 (28.89) | |
| 3 | 0 (0) | 13 (28.89) | |
| 4 | 0 (0) | 7 (15.56) | |
| 5 | 0 (0) | 12 (26.67) | |
| Is path DIL hemigland in the same hemigland as the biggest tumor on MRI | |||
| No | 1 (50) | 3 (7.89) | 0.19 |
| Yes | 1 (50) | 35 (92.11) | |
| Absolute difference in path DIL max dimension and PET max diameter in the same hemigland as prostatectomy | |||
| Agreement (≤5) | 0 (0) | 35 (77.78) | 0.06 |
| Disagreement (>5) | 2 (100) | 10 (22.22) | |
| Absolute difference in path dom nod max dimension and MRI max diameter in the same hemigland as prostatectomy | |||
| Agreement (≤5) | 2 (100) | 20 (55.56) | 0.50 |
| Disagreement (>5) | 0 (0) | 16 (44.44) | |
| Difference in path dom nod max dimension and PET max diameter in the same hemigland as prostatectomy | |||
| Max diameter < path dom nod max dimension | 0 (0) | 26 (59.09) | 0.50 |
| Max diameter > path dom nod max dimension | 0 (0) | 18 (40.91) | |
| Difference in path dom nod max dimension and MRI max diameter in the same hemigland as prostatectomy | |||
| Max diameter < path dom nod max dimension | 1 (50) | 22 (64.71) | 1.00 |
| Max diameter > path dom nod max dimension | 1 (50) | 12 (35.29) | |
Data are presented as the number of patients (column %), mean (± SD), or median (IQR). P value is calculated by analysis of variance or Wilcoxon rank-sum test for continuous and ordinal variables; and chi-square test or Fisher's exact test for categorical variables, where appropriate. Significant results are in bold.
Abbreviations: DIL = dominant intraprostatic lesion; MRI = magnetic resonance imaging; PET = positron emission tomography; PI-RADS = prostate imaging reporting and data system; PSA = prostate-specific antigen; SUV = standardized uptake value.
We also evaluated whether biopsy or PET findings could have aided in identifying the correct location of the DIL in the 4 cases where the MRI did not correspond to the DIL hemigland on final pathology. One of these cases was described above, as the patient who had a negative PET scan, biopsy findings with the same maximum Gleason score in both hemiglands, and an MRI lesion that was mismatched with the DIL. In this scenario, no modality would have provided a clear indication of the correct hemigland. However, for the remaining 3 cases (3/4), both the biopsy and the PET findings correctly matched the hemigland of the DIL, while the MRI was the only modality that did not.
Univariate and multivariable analysis of the absolute difference between imaging and pathologic diameter between MRI DIL and pathological DIL was performed when considering cliniopathologic factors that would be available at the time of diagnosis (Table 5). On univariate analysis, higher PSA was associated with a higher absolute difference between imaging and pathologic DIL diameter. On multivariable analysis, after adjusting for PI-RADS score, higher PSA remained associated with higher absolute differences between imaging and pathologic diameter.
Table 5.
Univariate and multivariable analyses of absolute difference between imaging and pathologic diameter on MRI with clinicopathologic characteristics
| Clinicopathologic characteristic | Univariate |
Multivariable* |
||
|---|---|---|---|---|
| Estimate (95% CI) | P value | Estimate (95% CI) | P value | |
| PSA† | 0.197 (0.139, 0.255) | <.001 | 0.344 (0.139, 0.550) | .002 |
| Gleason group on biopsy | ||||
| 1/2/3 | −0.299 (−3.178, 2.580) | .840 | ||
| 4/5 | 0 (reference) | |||
| PI-RADS score of the biggest MRI lesion in prostatectomy hemigland DIL† | −0.918 (−2.934, 1.098) | .378 | −1.846 (−4.030, 0.339) | .107 |
Abbreviation: CI = confidence interval; DIL = dominant intraprostatic lesion; MRI = magnetic resonance imaging; PI-RADS = prostate imaging reporting and data system; PSA = prostate-specific antigen.
Model was fit using a t family distribution with an identity link function including 37 observations.
Estimate is expressed as 1-unit increment.
Discussion
Microboosting the DIL is an emerging strategy in prostate cancer treatment with improved efficacy and similar toxicities to standard plans.5,12,22, 23, 24 However, significant challenges remain in the identification of the DIL, reproducible delineation of the DIL and defining appropriate CTV margins.13 In this study, we evaluated potential predictors of mismatch in hemigland location and size discrepancies among the largest diameter lesion identified on preoperative PSMA PET and mpMRI compared with the pathological findings on prostatectomy. These types of radiology pathology correlation studies are critical given that radiation oncologists treat with an organ preservation approach.
When partial matches were considered a match, the DIL on the final pathology was in the same hemigland as the maximum lesion on the PET scan in 94% of patients (45 of 48) and in 90% of patients (36 of 40) on mpMRI. In addition, there were only 2/45 (4%) cases in which there was a hemigland match on PSMA but the lesion location within the hemigland was not a match. In all MRI cases, the hemigland and the lesion location matched on final pathology. These findings are consistent with recent literature that suggests a high sensitivity and specificity for PSMA PET and mpMRI finding clinically significant intraprostatic disease.8,14 It is also in line with data supporting the notion that combining PSMA PET and mpMRI enhances lesion detection and localization. In the 3 cases where PSMA PET incorrectly identified the DIL hemigland, mpMRI would have been helpful in 2 cases (for the third patient, an mpMRI was not available). In a similar vein, we also examined the utility of PSMA PET in cases where the mpMRI was incorrect. Among the 4 patients in which MRI mislocalized the DIL, PSMA PET and biopsy findings correctly identified the hemigland in 3. These findings suggest that incorporating multiple imaging modalities and the biopsy location information may improve DIL identification for the 5% to 10% of cases where there may be a mismatch.
Although multivariable analysis was limited by the number of events, there are certain features on PSMA and mpMRI that may increase the risk of a mismatch. A lower biopsy grade group was associated with a greater mismatch in DIL size between PSMA and final pathology, while a higher PSA on both PSMA and mpMRI was associated with a higher risk of imaging and final pathology size discrepancy. These findings are consistent with previously reported studies on limitations with both PSMA and mpMRI demonstrating lower radiology pathology concordance with lower grade groups and higher PSA density.7,15, 16, 17
Separate from hemigland localization, we also evaluated size discrepancies in the maximum diameter of the DIL on imaging and final pathology. The median size discrepancy on PSMA and pathology was 3.4 mm (IQR, 1.7-6.1), 25% of cases had a discrepancy of more than 5 mm, and the size on PSMA imaging was smaller than the pathologic DIL size in about 0% of cases. The median size discrepancy on mpMRI and pathology was 4.0 mm (IQR, 3-9), 42% of cases had a discrepancy of more than 5 mm, and the size on mpMRI imaging was smaller than the pathologic DIL size in about 64% of cases. Related to this, about 25% of cases were actually only a partial match meaning that the correct hemigland was identified but on final pathology there was extension across the midline which was not identified on the imaging. A common margin used in DIL boost planning is to add 2 to 5 mm to the visible disease on imaging.18, 19, 20 Based on our study data, this standard and symmetrical margin expansion on the imaging-based gross tumor volume (GTV) is going to underdelineate and overdelineate the true extent of disease in a substantial proportion of patients. Our data set is too small to determine what factors are most associated with either an underestimation or overestimation of tumor size. Given the dose response seen in the FLAME trial to the gross target volume ensuring adequate coverage is critical.21 This study is also not able to further evaluate how targeted biopsy and artificial intelligence information could be used to develop personalized expansions on the imaging GTV.22 In fact, in the urology focal therapy literature, the concept of a penumbra (10 mm within a region of interest) has been defined demonstrating various amounts of disease outside of the MRI-defined lesion that depends on the grade of the lesion.25 The location of targeted biopsies (within target or perilesion) is not typically reported in pathology reports but perhaps this type of information would be helpful to radiation oncologists when applying margins to an imaging-based GTV. Despite these limitations in boosting the “ground truth,” the results of FLAME and other microboost studies do demonstrate improved outcomes even with our current approaches.15
The main limitation of our study was the limited sample size, which constrained the statistical power of our analyses and limited our ability to perform multivariable analysis. Future studies with larger data sets are essential to confirm our findings and better identify predictors of location mismatch and tumor dimension agreement/discrepancies between mpMRI, PSMA PET, and final pathology. We also did not explore changes in lesion diameter measurements on PSMA PET based on different SUV thresholds (ie, 30%, 40%, SUVbackground × 2). Some work suggests that the lesion diameter on PSMA is closer to the true pathologic diameter using different SUV thresholds; however, there are still substantial differences between diameters and there is no agreed-upon thresholding at this time.26, 27, 28 Some of this variability may be related to challenges in lesions with low SUVmax as well as larger lesions and challenges of accurately determining background SUV. In a recent trial that includes an MRI and PSMA-based focal boost, a standard SUV threshold was not set and they used variable SUV thresholds to try to match the MRI-based DIL volume. Even when doing this, their median SUV threshold was 48% but there was an IQR of 40% to 58%. This raises the complexities with trying to standardize the SUV threshold to use when contouring a DIL.18 Similarly, we only looked at a maximum diameter and did not evaluate GTV volumes. Finally, there was a lack of standardized imaging practice; however, we believe that this is reflective of what happens in a real-world practice setting. Imaging was not standardized to 1 location, and there was a notable discrepancy in the time interval between imaging and prostatectomy. The median (IQR) time between PET and prostatectomy was 43.5 (21.5-69) days, and between MRI and prostatectomy was 108 (66-176) days. While time interval was not found to be a statistically significant factor in hemigland concordance or size discrepancy in our study, the variability in timing may limit the accuracy of direct comparisons between PET and MRI.
The long-term goal of this work is to develop a standardized framework for using PSMA PET/CT and mpMRI to guide microboosting of prostate tumors, including the determination of appropriate margins around the GTV. Consistent with other work, both imaging modalities appear complementary to one another. Understanding size discrepancies between imaging and final pathology needs further investigation and perhaps even some type of nomogram to estimate the risks of overestimation versus underestimation.
Conclusions
Our study suggests that factors associated with smaller, less aggressive tumors may reduce the sensitivity of both PSMA PET/CT and mpMRI. These findings highlight the importance of identifying and validating factors that may influence margins needed for effective microboosting.
Disclosures
Royalties or licenses—Springer book royalties; consulting fees—Mdxhealth; Advisory board – Merk; payment for honoraria for lectures, presentations, speakers, bureaus, manuscript writing or educational events—UroToday; participation on a data safety monitoring board or advisory board—CHUM DSMB, Alessa, GT Medical; leadership or fiduciary role in other board, society, committee or advocacy group, paid or unpaid—GU ASCO Steering Committee, ASTRO scientific session vice chair_GU, ASTRO board of directors, ABS President.
Acknowledgments
Sungjin Kim was responsible for statistical analysis.
Footnotes
Sources of support: This work had no specific funding.
Research data are stored in an institutional repository and will be shared on request to the corresponding author.
References
- 1.Ma T.M., Chu F.I., Sandler H., et al. Local failure events in prostate cancer treated with radiotherapy: a pooled analysis of 18 randomized trials from the meta-analysis of randomized trials in cancer of the prostate consortium (LEVIATHAN) Eur Urol. 2022;82:487–498. doi: 10.1016/j.eururo.2022.07.011. [DOI] [PubMed] [Google Scholar]
- 2.Smith C.P., Xiang M., Armstrong W.R., et al. Patterns of failure in men with radiorecurrent prostate cancer: a post hoc analysis of 3 prospective gallium 68 prostate-specific membrane antigen positron emission tomography/computed tomography imaging trials. Int J Radiat Oncol Biol Phys. 2023;116:1079–1084. doi: 10.1016/j.ijrobp.2023.02.039. [DOI] [PubMed] [Google Scholar]
- 3.Aizawa R., Otani T., Ogata T., et al. Spatial pattern of intraprostatic recurrence after definitive external-beam radiation therapy for prostate cancer: implications for focal boost to intraprostatic dominant lesion. Adv Radiat Oncol. 2024;9 doi: 10.1016/j.adro.2024.101489. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Goodman C.D., Fakir H., Pautler S., Chin J., Bauman GS. Dosimetric evaluation of PSMA PET-delineated dominant intraprostatic lesion simultaneous infield boosts. Adv Radiat Oncol. 2019;5:212–220. doi: 10.1016/j.adro.2019.09.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Zhong A.Y., Lui A.J., Katz M.S., et al. Use of focal radiotherapy boost for prostate cancer: radiation oncologists' perspectives and perceived barriers to implementation. Radiat Oncol. 2023;18:188. doi: 10.1186/s13014-023-02375-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Zhong A.Y., Lui A.J., Kuznetsova S., et al. Clinical impact of contouring variability for prostate cancer tumor boost. Int J Radiat Oncol Biol Phys. 2024;120:1024–1031. doi: 10.1016/j.ijrobp.2024.06.007. [DOI] [PubMed] [Google Scholar]
- 7.Westphalen A.C., McCulloch C.E., Anaokar J.M., et al. Variability of the positive predictive value of PI-RADS for prostate MRI across 26 centers: experience of the society of abdominal radiology prostate cancer disease-focused panel. Radiology. 2020;296:76–84. doi: 10.1148/radiol.2020190646. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Mazzone E., Cannoletta D., Quarta L., et al. A comprehensive systematic review and meta-analysis of the role of prostate-specific membrane antigen positron emission tomography for prostate cancer diagnosis and primary staging before definitive treatment. Eur Urol. 2025;87:654–671. doi: 10.1016/j.eururo.2025.03.003. [DOI] [PubMed] [Google Scholar]
- 9.Rigby R.A., Stasinopoulos DM. Generalized additive models for location, scale and shape. J R Stat Soc C Appl Stat. 2005;54:507–544. [Google Scholar]
- 10.van Buuren S., Fredriks M. Worm plot: simple diagnostic device for modelling growth reference curves. Stat Med. 2001;20:1259–1277. doi: 10.1002/sim.746. [DOI] [PubMed] [Google Scholar]
- 11.R Core Team . R Foundation for Statistical Computing; 2025. R: A language and environment for statistical computing. [Google Scholar]
- 12.Kerkmeijer L.G.W., Groen V.H., Pos F.J., et al. Focal boost to the intraprostatic tumor in external beam radiotherapy for patients with localized prostate cancer: results from the FLAME randomized phase III trial. J Clin Oncol. 2021;39:787–796. doi: 10.1200/JCO.20.02873. [DOI] [PubMed] [Google Scholar]
- 13.Patel K.R., van der Heide U.A., Kerkmeijer L.G.W., et al. Target volume optimization for localized prostate cancer. Pract Radiat Oncol. 2024;14:522–540. doi: 10.1016/j.prro.2024.06.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Emmett L., Buteau J., Papa N., et al. The additive diagnostic value of prostate-specific membrane antigen positron emission tomography computed tomography to multiparametric magnetic resonance imaging triage in the diagnosis of prostate cancer (PRIMARY): a prospective multicentre study. Eur Urol. 2021;80:682–689. doi: 10.1016/j.eururo.2021.08.002. [DOI] [PubMed] [Google Scholar]
- 15.Sonni I., Felker E.R., Lenis A.T., et al. Head-to-head comparison of 68Ga-PSMA-11 PET/CT and mpMRI with a histopathology gold standard in the detection, intraprostatic localization, and determination of local extension of primary prostate cancer: results from a prospective single-center imaging trial. J Nucl Med. 2022;63:847–854. doi: 10.2967/jnumed.121.262398. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Bagheri H., Mahdavi S.R., Geramifar P., et al. An update on the role of mpMRI and 68Ga-PSMA PET imaging in primary and recurrent prostate cancer. Clin Genitourin Cancer. 2024;22 doi: 10.1016/j.clgc.2024.102076. [DOI] [PubMed] [Google Scholar]
- 17.Pooli A., Johnson D.C., Shirk J., et al. Predicting pathological tumor size in prostate cancer based on multiparametric prostate magnetic resonance imaging and preoperative findings. J Urol. 2021;205:444–451. doi: 10.1097/JU.0000000000001389. [DOI] [PubMed] [Google Scholar]
- 18.Singh M., Katdare A., Ghosh S., et al. PSMA-PET guided intraprostatic boost in prostate SBRT (PROBE): a phase 2 trial. Int J Radiat Oncol Biol Phys. 2025 doi: 10.1016/j.ijrobp.2025.03.033. [DOI] [PubMed] [Google Scholar]
- 19.Spohn S.K.B., Zamboglou C., Bürkle S.L., et al. Safety and quality of life of PSMA-PET- and MRI-based focal dose escalated radiotherapy for intermediate- and high-risk prostate cancer: primary endpoint analysis of the bi-centric phase II HypoFocal trial (ARO2020-01) Radiother Oncol. 2025;208 doi: 10.1016/j.radonc.2025.110883. [DOI] [PubMed] [Google Scholar]
- 20.Draulans C., Haustermans K., Pos F.J., et al. Stereotactic body radiotherapy with a focal boost to the intraprostatic tumor for intermediate and high risk prostate cancer: 5-year efficacy and toxicity in the hypo-FLAME trial. Radiother Oncol. 2024;201 doi: 10.1016/j.radonc.2024.110568. [DOI] [PubMed] [Google Scholar]
- 21.Menne Guricová K., Pos F.J., Schoots I.G., et al. Intra-prostatic recurrences after radiotherapy with focal boost: location and dose mapping in the FLAME trial. Radiother Oncol. 2024;201 doi: 10.1016/j.radonc.2024.110535. [DOI] [PubMed] [Google Scholar]
- 22.Mouraviev V., Villers A., Bostwick D.G., Wheeler T.M., Montironi R., Polascik TJ. Understanding the pathological features of focality, grade and tumour volume of early-stage prostate cancer as a foundation for parenchyma-sparing prostate cancer therapies: active surveillance and focal targeted therapy. BJU Int. 2011;108:1074–1085. doi: 10.1111/j.1464-410X.2010.10039.x. [DOI] [PubMed] [Google Scholar]
- 23.Pucar D., Hricak H., Shukla-Dave A., et al. Clinically significant prostate cancer local recurrence after radiation therapy occurs at the site of primary tumor: magnetic resonance imaging and step-section pathology evidence. Int J Radiat Oncol Biol Phys. 2007;69:62–69. doi: 10.1016/j.ijrobp.2007.03.065. [DOI] [PubMed] [Google Scholar]
- 24.von Eyben F.E., Kiljunen T., Kangasmaki A., Kairemo K., von Eyben R., Joensuu T. Radiotherapy boost for the dominant intraprostatic cancer lesion-A systematic review and meta-analysis. Clin Genitourin Cancer. 2016;14:189–197. doi: 10.1016/j.clgc.2015.12.005. [DOI] [PubMed] [Google Scholar]
- 25.Brisbane W.G., Priester A.M., Ballon J., et al. Targeted prostate biopsy: umbra, penumbra, and value of perilesional sampling. Eur Urol. 2022;82:303–310. doi: 10.1016/j.eururo.2022.01.008. [DOI] [PubMed] [Google Scholar]
- 26.Kivikallio A., Malaspina S., Saarinen I., et al. Prospective comparison of 18F-PSMA-1007 PET/CT and MRI with histopathology as the reference standard for intraprostatic tumour detection and T-staging of high-risk prostate cancer. Eur J Nucl Med Mol Imaging. 2025;52:3709–3719. doi: 10.1007/s00259-025-07208-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Spohn S.K.B., Kramer M., Kiefer S., et al. Comparison of manual and semi-automatic [18F]PSMA-1007 PET based contouring techniques for intraprostatic tumor delineation in patients with primary prostate cancer and validation with histopathology as standard of reference. Front Oncol. 2020;10 doi: 10.3389/fonc.2020.600690. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Draulans C., De Roover R., van der Heide U.A., et al. Optimal 68Ga-PSMA and 18F-PSMA PET window levelling for gross tumour volume delineation in primary prostate cancer. Eur J Nucl Med Mol Imaging. 2021;48:1211–1218. doi: 10.1007/s00259-020-05059-4. [DOI] [PubMed] [Google Scholar]