Using MRI Apparent Diffusion Coefficient (ADC) Values to Reduce Over-Diagnosis and Over-Biopsy of Suspected Prostate Cancer

Introduction

Over-diagnosis of prostate cancer remains a clinical concern, often leading to unnecessary biopsies and anxiety for patients with low-grade, clinically insignificant (CI) disease. Imaging parameters, such as the Apparent Diffusion Coefficient (ADC) from diffusion-weighted MRI could be utilised to better detect clinically significant (CS) prostate cancer prior to biopsy.

Methods

This retrospective study included patients referred for suspected prostate cancer at Leeds Teaching Hospitals NHS Trust between January 2021 and June 2025. Patients who proceeded to a biopsy were classified as either no cancer, clinically insignificant (CI) cancer (ISUP grade ≤ 2) and clinically significant (CS) cancer (ISUP grade ≥ 3). ADC values were recorded for patients with a suspicious focal lesion on MRI. Diagnostic performance was assessed across ADC thresholds corresponding to the 95^th and 99^th percentile for CS cancer.

Results

A total of 4802 patients underwent MRI as part of their diagnostic pathway. Of these N=1540/4802 (32.0%) had a suspicious focal lesion that was given an ADC reading. Of these N=1498/1540 (97.3%) proceeded to a biopsy. Of the patients with an ADC reading and biopsy results (4 patients did not have an ISUP grade), N=267/1494 (17.9%) of patients had no cancer, N=693/1494 (46.4%) had CI cancer, and N=534/1494 (35.7%) had CS cancer.

For CS cancer, the 95^th and 99^th percentile of ADC Focal values fell at 947 and 1192 x 10^-6 mm²/s. If no patients with an ADC of >947 had been offered a biopsy, N=180/267 (67.4%) of patients with no cancer and N=155/693 (22.4%) of patients with CI disease would have been spared a biopsy and n=27/534 (5.1%) of patients with CS disease would have been missed. If no patients with an ADC of >1192 had been offered a biopsy, N=66/267 (24.7%) of patients with no cancer and N=42/693 (6.1%) of patients with CI disease would have been spared a biopsy and N=6/534 (1.1%) of patients with CS disease would have been missed.

Discussion and Conclusion

Higher ADC values of suspicious focal lesions on MRI were associated with a lower likelihood of clinically significant prostate cancer. Using ADC thresholds of 95^th and 99^th percentile to filter out patients from going for biopsy would have had the benefit of saving 67% or 25% of current patients with no cancer and 22% or 6% of CI disease patients from going for a biopsy, with the cost of 5% and 1% of CS cancers being missed.

Limitations include the ADC value coming from an elliptical tool, rather than whole focal lesion evaluation. ADC values were only given when there was a presence of a suspicious area on MRI.