Multiparametric MRI often underestimates pathological tumor size in prostate cancer
Click Here to Manage Email Alerts
Multiparametric MRI often underestimated pathological tumor size among men with biopsy-proven prostate cancer, according to single-arm study results published in Journal of Urology.
The extent of underestimation often increased with smaller radiologic tumor size and lower Prostate Imaging Reporting and Data System (PI-RADS) version 2 scores.
Consequently, these variables must be taken into account when estimating treatment margins, according to first author Aydin Pooli, MD, urologic oncologist and health sciences clinical instructor at UCLA Jonsson Comprehensive Cancer Center and David Geffen School of Medicine at UCLA, and colleagues.
“Aside from careful patient selection, one of the key aspects of successful focal ablation therapy includes accurate delineation of the tumor margin,” Pooli told Healio. “This is a complex procedure, especially due to challenges in co-registration and potential pitfalls in spatial resolution and sectioning geometrics.
“Urologists performing focal therapy cannot assume they can fully compensate for any radiologic underestimation of tumor size with literal margin expansion of the magnetic resonance target [because] the tumor shape is not always spherical or ovoid,” Pooli added. “It appears that multiparametric MRI can be inaccurate in terms of delineating the prostate cancer index lesion. Therefore, in order to ablate the prostate tumor effectively, we should first develop better techniques to determine the shape, size and margins of the tumor.”
Radical prostatectomy or external beam radiation therapy with or without androgen deprivation therapy is standard for men with localized intermediate- or high-risk prostate cancer, as well as for some men with low-risk disease.
“However, recently, focal ablative therapies with either high-intensity focused ultrasound or focal laser ablation therapy are gaining interest and marketed as minimally invasive but effective therapies for prostate cancer,” Pooli said. “These focal ablative therapies are usually based on multiparametric prostate MRI to determine the tumor margins.”
The single-arm cohort study — led by Robert E. Reiter, MD, professor of urology, director of urologic research and director of the prostate cancer program at UCLA — assessed agreement between radiologic tumor size and true tumor size after surgical excision based on whole-mount histopathology slides. Researchers also aimed to identify predictors of pathological tumor size.
The cohort included 441 consecutive men with biopsy-proven prostate cancer and PI-RADS scores of 3 or greater on multiparametric MRI. All men subsequently underwent radical prostatectomy.
Researchers assessed the difference between radiologic and pathological tumor sizes of 461 lesions. They calculated mean radiologic tumor size of 1.57 cm and mean pathological tumor size of 2.37 cm (P < .001).
Results showed radiologic tumor size consistently underestimated pathological tumor size. The extent of underestimation increased with smaller radiologic tumor size.
Multiparametric MRI underestimated true tumor size by an average of 1.25 cm (median, 0.9; standard deviation [SD], 0.93) for lesions smaller than 1 cm on multiparametric MRI, 0.86 cm (median, 0.7; SD, 0.77) for lesions 1 cm to 2 cm on MRI, and 0.3 cm (median, 0.3; SD, 1.02) for lesions greater than 2 cm on MRI.
“In other words, the true tumor size was up to 3.11 cm larger than radiologic tumor size for lesions smaller than 1 cm on multiparametric prostate MRI, up to 2.4 cm larger than radiologic tumor size for lesions 1 cm to 2 cm, and ... up to 2.34 cm larger than radiologic tumor size for lesions greater than 2 cm on multiparametric prostate MRI,” Pooli said.
The degree of underestimation was slightly smaller for lesions with PI-RADS scores of 5, Pooli said.
“The overall correlation between radiologic tumor size and pathological tumor size was poor, and inclusion of other variables — including biopsy Gleason grade, tumor location, zone and PI-RADS categories — did not improve the prediction model for true tumor size,” Pooli said.
Pathological tumor size was significantly larger in cases with biopsy Gleason Grade Group 5 compared with Gleason Grade Group 1 (mean change, 0.37 cm; P = .014), PI-RADS 5 lesions compared with PI-RADS 4 lesions (mean change, 0.26; P = .006) and higher PSA density.
“Focal ablative therapy for prostate cancer has generally been used for smaller tumor size in multiparametric MRI, and for tumors with lower suspicious scores on multiparametric prostate MRI,” Pooli said. “Our study demonstrated that the degree of underestimation of true tumor size is more pronounced for smaller tumor size.
“In other words, in order to ablate a tumor with radiologic tumor size less than 1 cm on MRI with 97.5% certainty, a treatment margin of 3.11 cm — corresponding to the 97.5th percentile of the predictions — would be required based on assumed distribution, which is a very large margin and may be larger than the prostate size at times,” he added. “Additionally, the degree of underestimation for lesions with lower suspicion score on prostate MRI was more pronounced than for the ones with high suspicion score (PI-RADS 5 vs PI-RADS 2).”
The incorporation of tracked biopsy around the multiparametric prostate MRI-detected lesion may help overcome the limitations of MRI for characterizing tumor size, Pooli said.
“In association with our radiology colleagues, we are in the process of hypothesis development to design a study using tracked biopsy to determine the shape of the tumor,” he told Healio. “Further work in this area should seek to develop algorithms that accurately combine radiologic tumor size and patient-specific factors to correctly predict pathological tumor size.
“Although our data suggest there is an underestimation of pathological tumor size by radiologic tumor size, even after combining this data with patient- and disease-specific factors, we were unable to predict the exact extent of underestimation on a case-by-case basis,” Pooli added. “Until these algorithms exist, care must be taken when both interpreting and using multiparametric prostate MRI for treatment selection and planning.”
Collapse
Click Here to Manage Email Alerts