Determination of the optimal cut-off value of serum prostate-specific antigen in the prediction of skeletal metastases on technetium-99m whole-body bone scan by receiver operating characteristic curve analysis

Enhancing bone metastasis prediction in prostate cancer using quantitative mpMRI features, ISUP grade and PSA density: a machine learning approach

PURPOSE

Bone metastasis is a critical complication in prostate cancer, significantly impacting patient prognosis and quality of life. This study aims to enhance bone metastasis prediction using machine learning (ML) techniques by integrating dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) perfusion features, International Society of Urological Pathology (ISUP) grade, and prostate-specific antigen (PSA) density.

MATERIALS AND METHODS

A retrospective analysis was conducted on 122 patients with histopathologically confirmed prostate cancer who underwent multiparametric prostate magnetic resonance imaging (mpMRI). Quantitative mpMRI features, PSA density, and ISUP grades were extracted and normalized. The dataset was balanced using oversampling and divided into training (70%) and test (30%) sets. Various ML models were developed and evaluated using area under the curve (AUC) metrics.

RESULTS

Bone metastases were present in 26 patients (21.3%) at diagnosis. IAUGC and MaxSlope showed a statistically significant association with bone metastasis (p = 0.035, p = 0.050 respectively). The optimal PSA density cut-off value of 0.24 yielded a sensitivity of 0.88, specificity of 0.60, and AUC of 0.77. Machine learning models were developed using the dataset created with IAUGC, MaxSlope, ISUP grade, and PSA density values. Among the ML models, XGBoost demonstrated superior performance with validation and test AUCs of 91.5% and 92.6%, respectively, along with high precision (93.3%) and recall (93.1%).

CONCLUSION

Integrating quantitative mpMRI features, ISUP grade, and PSA density through machine learning algorithms, particularly XGBoost, significantly improves the accuracy of bone metastasis prediction in prostate cancer patients. This approach can potentially reduce the need for additional imaging modalities and associated radiation exposure.

Determination of the optimal imaging protocol for [18F]PSMA-PET-CT for the detection of bone metastases in prostate cancer patients

AIM

Prostate-specific membrane antigen-positron emission tomography (PSMA-PET) is a widely used diagnostic tool in patients with prostate cancer (PC). However, due to the limited availability of PET scanners and relevant acquisition costs, it is important to consider the indications and acquisition time. The aim of this investigation was to determine whether a PET scan from the skull base to the proximal thigh is sufficient to detect the presence of bone metastases.

METHODS

A retrospective analysis was conducted on 1050 consecutive [18F]PSMA-1007-PET-CT scans from the head to the proximal lower leg. The PET scans were categorised according to the presence and amount of bone metastases: (1) 1-5, (2) 6-19 and (3) ≥20. Additionally, the PET scans were evaluated for the presence of bone metastases below the proximal thigh as well as bone metastases above the skull base. Imaging results were compared to patients PSA values.

RESULTS

Of the 391 patients with bone metastases, 146 (37.3%) exhibited metastases located below the proximal thigh and 104 (26.6%) above the skull base. The majority of bone metastases located below the proximal thigh (145, 99.3%) and above the skull base (94, 90.4%) were identified in patients with more than five bone metastases. No solitary distal metastasis was detected. The PSA value correlated significantly with number of bone metastases (e. g., 1-5 vs. ≥20 bone metastases, P < 0.001) and was significantly higher in patients with distal bone metastases (P < 0.001). ROC analysis showed that a PSA value of 11.15 ng/mL is the optimal cut-off for detecting bone metastases located below the proximal thigh, with an AUC of 0.919 (95% CI: 0.892-0.945, sensitivity 87%, specificity 86%). Similarly, the PSA value of 12.86 ng/mL is the optimal cut-off for detecting bone metastases above the skull base with an AUC of 0.904 (95% CI: 0.874-0.935, sensitivity 87%, specificity 83%).  CONCLUSION: PSMA-PET acquisition protocols from the skull base to the proximal femur may be sufficient to accurately detect bone metastatic disease in PC. PSA values can provide decision support for individual PET acquisition protocols.

Diagnostic accuracy of bone scan at different PSA levels in biochemical recurrence of prostate cancer

OBJECTIVE

To determine the diagnostic accuracy of Bone Scan at different PSA levels for detecting skeletal metastases in men with biochemical recurrence of prostate cancer.

METHODS

We conducted a retrospective review of the statewide RIS-PACS to identify 251 men with biochemical recurrence who underwent both a Bone Scan and Ga68 PSMA PET/CT (within 2 months of each other) between September 2019 and December 2022 at a single institution. The Ga68 PSMA PET/CT report was considered to be the reference standard.

RESULTS

The median age was 72 years (IQR 67-76) with a median PSA level of 1 ng/ml (IQR 0.25-2.8). Using Ga68 PSMA PET/CT as the reference standard, 68/251 patients (25%) were positive for osseus metastases. Overall sensitivity and specificity of Bone Scan was 51% (95% CI 40-64%) and 99% (95% CI 98-100%) respectively. Using PSA banding, a PSA threshold of 20 ng/ml provided the greatest discriminatory benefit with sensitivity of the Bone Scan below the threshold being 46% (95% CI 33-59%) and above the threshold being 89% (95% CI 68-100%). Specificity remained consistently high both below and above this threshold.

CONCLUSION

Bone Scan provides greater diagnostic accuracy for detecting skeletal metastases in biochemical recurrence when the PSA level is above 20 ng/ml. This knowledge is valuable in optimising imaging algorithms in biochemical recurrence, particularly in regions where PSMA PET/CT is less readily available or affordable.

Segmented linear correlations between bone scan index and prostate cancer biomarkers, alkaline phosphatase, and prostate specific antigen in patients with a Gleason score ≥7

Technetium-99m methyl diphosphonate bone scintigraphy is relatively easily accessible for detecting bone metastases in prostate cancer patients. However, it is subjective and can be challenging to compare images taken at different time points. The bone scan index (BSI) is a more objective evaluation and allows for better comparison of images. Its correlation with other biomarkers of prostate cancer metastases such as prostate specific antigen (PSA) and alkaline phosphatase (ALP) is not clearly understood. This study thus aimed to compare the BSI correlation to PSA against that of BSI to ALP levels in patients with a Gleason score ≥7.A retrospective analysis of the medical records of 50 prostate cancer patients with a Gleason score of ≥7 referred for a bone scan between January 1, 2015 and December 31, 2018 was undertaken. Bone scans were interpreted visually, and using a semi-automated computer programme to quantify the BSI and its relation to PSA and ALP measurements.For the metastasis positive measurements, there was a statistically significant moderate positive overall linear correlation between BSI and PSA. For ALP and BSI, there were 2 segmented strong positive linear relationships between them. The first segment consisted of ALP < 375 IU/L and BSI >10%, where ALP and BSI were strongly and positively correlated. The other segment tended to have generally low BSI measurements (<10%) and also had a strong and positive correlation.The BSI was found to be better linearly correlated with ALP than PSA.

False Positive Findings of [18F]PSMA-1007 PET/CT in Patients After Radical Prostatectomy with Undetectable Serum PSA Levels

Background

PET-CT using prostate-specific membrane antigen (PSMA)-targeting radiopharmaceuticals labeled with ⁶⁸Ga or ¹⁸F has emerged as the most sensitive staging tool in prostate cancer (PC). Nonetheless, the occurrence of false positive (FP) findings presents a major concern of this approach. In this prospective study, we investigated the frequency and pattern of false-positive findings of [¹⁸F]PSMA-1007 PET/CT in patients after radical prostatectomy with undetectable serum PSA levels. Any discrete non-physiological accumulation of [¹⁸F]PSMA-1007 in this population is by definition FP.

Methods

Seventeen men after radical prostatectomy, whose serum PSA levels were <0.05 ng/mL at 2–24 months after surgery were prospectively recruited. PET/CT was acquired at both 1 and 2 h after injection of [¹⁸F]PSMA-1007.

Findings

Three studies (18%) were interpreted as completely normal. Thirty-five foci of “non-physiological” uptake were observed in the remaining 14 (82%) patients, including a single skeletal focus in four patients, multiple skeletal foci in five patients and soft tissue uptake in eight, including in a desmoid tumor and in pelvic lymphocele. The SUV_max of all lesions was in the range of 1–7, except for the desmoid tumor which measured 12.7. All foci were visible in both the 1- and the 2 h studies, presenting a minor (<10%), statistically insignificant increase of SUV_max during this time-interval.

Interpretation

FP [¹⁸F]PSMA-1007-avid foci are found in about 80% of patients with undetectable serum PSA levels. Thus, focal uptake of [¹⁸F]PSMA-1007 outside its physiological distribution is not a categorical sign of metastasis and can arise from non-specific uptake of the ligand. The interpretation of [¹⁸F]PSMA-1007 PET/CT studies should always consider the clinical context, and lesions with SUV_max < 7 are suspicious for FP.