Pattern and Distribution of Skeletal Metastases in Patients With Prostate Cancer in Ghana: A Descriptive Analysis and a Model-Based Digital Nomogram for Oligo-Ostotic Versus Polyostotic Metastases.

backgroundBone involvement is a frequent and serious consequence of advanced prostate cancer. Understanding the distribution and pattern of bone involvement is essential for early detection, staging, and appropriate therapeutic intervention, particularly in low-resource settings where late presentation is frequent.  Objective: To characterize the anatomical distribution and burden of skeletal metastases in a cohort of newly diagnosed patients with prostate cancer with bone involvement.

methodsA one-year cross-sectional review of bone scans from 100 newly diagnosed patients with prostate cancer with confirmed skeletal metastases was conducted. Standardized site coding produced frequency tables, total/per-patient counts, and descriptive statistics (mean, median, mode, and IQR). Axial vs. appendicular patterns were shown with bar charts and skeletal heatmaps. A logistic regression model-girded digital nomogram (derived, validated, tested, and updated) predicted superscan/polyostotic metastasis risk. Analyses were in Stata Version 17 (StataCorp LLC, College Station, TX) and Python (Python Software Foundation, Fredericksburg, VA) at α = 0.05.

resultsA total of 470 bone metastatic sites were documented among 100 newly diagnosed patients with prostate cancer, aged 51 to 89 (mean age = 68.81, mode = 65, SD = 7.07 years) with a mean prostate-specific antigen (PSA) of 924.32 ng/ml (range = 5.76 to 2223, SD = 2656.47 ng/ml), alkaline phosphatase (ALP), mean = 239.43, range = 45 to 3265, SD = 438.57, digital rectal examination (DRE) risk (median = 3, IQR = 1), International Society of Urological Pathology (ISUP) risk (median = 3, IQR = 1), and D'Amico risk categories (median = 3, IQR = 0). The number of metastases per patient ranged from 1 to 19, with a mean of 4.7 sites (SD = 2.98), a median of 5 (IQR: 3-7), and a mode of 3. Across the 15 anatomic sites identified, the lesion per-anatomic site per-patient (mean, median, mode, and SD) were 0.31, 1.0, 3.0, and 2.55, respectively. There was notable variability in metastatic burden (solitary lesions in 33.0%, three or more lesions in 51.0%, and superscan/polyostotic lesions (> 4 metastatic sites) in 33.0%), also suggesting heavy burden disease amongst the cohort. The most commonly involved sites were the spine (32.6%), ribs (25.7%), and pelvis (16.1%), the skull (5.2%), and hand and foot bones had zero metastasis. The axial skeleton accounted for 68.1% of all metastatic deposits, and this centripetal spread was further highlighted by heatmap visualizations. Phenotypic clustering revealed that a patient with PSA > 100 ng/ml but moderate ALP (62-122 U/L) is most likely to fall in Q1 (oligo-metastatic pattern; likelihood-ratio (LR) = 31.33 (df = 4), p < 0.0001). Conversely, the combination of age (55-75 years), PSA > 100 ng/ml, and ALP ≥ 222 U/L sharply elevates the likelihood of Q3 (polyostotic disease), regardless of already-high clinical risk scores (LR = 29.77 (df = 4), p < 0.0001). A logistic regression model-girded digital nomogram for oligo-ostotic versus polyostotic metastatic prostate cancer was derived, validated, trained, and deployed, achieving an area under the curve (AUC) of 81.0% at the end.

conclusionAxial skeletal metastasis dominates in prostate cancer bone metastasis. Heavy burden metastasis was rampant. These findings emphasize the need for early diagnostic strategies and risk-adapted imaging protocols to improve outcomes in resource-limited settings.