The role of imaging in a patient-tailored approach for prostate cancer

Publication date: 2013-06


Van den Bergh, Laura


Prostate cancer is the most common non-skin malignancy in men in developed countries and the third leading cause of cancer-related mortality. Notwithstanding continuous improvements in the area of surgery and radiotherapy, the two most important treatment modalities for locoregional disease, there are still patients in whom definite tumor control cannot be achieved. One of the obstacles for the adequate treatment of prostate cancer is that neither nomograms nor conventional imaging techniques allow us to reliably predict lymph node status in newly diagnosed patients. Therefore, an extended lymph node dissection is currently recommended as the only accurate option to assess nodal involvement. Since nodal status is crucial for treatment selection, there is an urgent need to investigate other strategies for lymph node staging. Next to nodal involvement, isolated local failure is also responsible for part of the biochemical failures. In patients scheduled for external beam radiotherapy, focal dose-escalation has been hypothesized to improve the therapeutic ratio without increasing toxicity rates as observed when escalating the dose to the entire gland. However, a prerequisite for this focal boosting strategy is accurate detection and localization of these intraprostatic lesions. With both problems in mind, we focused on the role of functional imaging in this project, to eventually evolve to a more patient-tailored treatment for prostate cancer. In order to achieve our goals, we performed a prospective clinical study, enrolling patients with localized or locally advanced, biopsy-proven invasive adenocarcinoma of the prostate, who were selected for surgical treatment. The most important inclusion criteria were a risk of finding affected lymph nodes of at least 10% but less than 35% according to the Partin tables and no evidence for the presence of lymph node metastases on computed tomography (CT). All patients underwent carbon-11 labeled choline positron emission tomography (PET)-CT and multiparametric magnetic resonance imaging (MRI) prior to surgery. On the day of the surgery, a sentinel node procedure was performed for which radiocolloïd was intraprostatically injected followed by imaging to allow localization of the sentinel nodes during surgery. Surgery consisted of the detection and resection of the sentinel lymph nodes and consecutive superextended lymphadenectomy (i.e. resection of lymph nodes in the obturator fossa, the internal, external and common iliac and presacral region), followed by radical prostatectomy. Lymph nodes were completely serially sectioned and meticulously microscopically examined after pankeratin staining. Radical prostatectomy specimens were processed according to the guidelines and cancerous regions within the prostate were delineated. Histopathology served as the gold standard for comparison with the imaging results. In total, 75 patients were enrolled in this study with assessment of the sensitivity, specificity, negative and positive predictive value of choline PET-CT and diffusion-weighted MRI as a primary endpoint. Thorough histopathological examination revealed the presence of lymph node metastases in no less than half of the patients. Unfortunately, the sensitivity of both imaging modalities appeared to be very low in the region-based analysis as well as in the patient-based analysis. Furthermore, a secondary analysis was performed only taking into account lymph node metastases larger than 2 mm, but this did not significantly change the predictive performance. Based on these data, 11C-choline PET-CT nor diffusion-weighted MRI could offer an alternative for the extended lymph node dissection for nodal staging in prostate cancer. The next goal of this project was to evaluate the feasibility and accuracy of the sentinel node procedure in prostate cancer. In total, 74 patients underwent the procedure. Apart from three patients, at least one sentinel node was detected and removed during surgery in all patients. The distribution of sentinel lymph nodes, the number as well as the localization, was highly variable between patients. Although we found the technique to be feasible in clinical practice, we considered the number of falsely negative staged patients to be too high to outweigh the invasiveness of this procedure in daily practice. With this project we also gained new insights in dissemination patterns of nodal metastases in prostate cancer. Analysis of node-positive patients learned that with the standard extended lymph node dissection template (i.e. external iliac and internal iliac region and obturator fossa), the vast majority of the patients was correctly staged as N1 but a number of affected lymph nodes would have been left behind. We could also observe a certain predilection for the internal iliac region and we demonstrated that the importance of the presacral region is currently underestimated. To maximize the chances for correct nodal staging and the chances to have removed all cancerous lymph nodes, we therefore recommend adding the presacral region to the standard extended template. From the radiation oncologist’s point of view, we could conclude that almost all retrieved malignant nodes would have been adequately covered with the currently used pelvic clinical target volumes. In the second part of this project, we focused on assessing the role of functional imaging for detection and localization of intraprostatic tumor nodules. In order to compare 11C-choline PET-CT, dynamic contrast-enhanced and diffusion-weighted images with the whole-mount prostatectomy sections, the prostate was divided into 24 segments per patient (basal, middle and apical part with 8 segments each). Suspicious regions were delineated on the prostatectomy sections as well as on the MRI images and a maximal standardized uptake value (SUVmax) was calculated per segment for choline PET-CT images. In total, 73 patients were available for analysis. Sensitivity and specificity varied as a function of the selected SUVmax threshold, and when selecting the cutoff with the highest accuracy, a relatively low sensitivity but high specificity was found. Then, T2-weigthed and functional MRI images were compared with histopathology results. The highest sensitivity for localization of intraprostatic lesions was obtained with diffusion-weighted MRI and further improved when combined with T2-weighted and dynamic contrast-enhanced MRI. Since the combination of imaging modalities frequently results into improved performance, we specifically aimed at assessing the potential additional value of 11C-choline PET-CT next to (functional) MRI. Unfortunately, we found its added value for detection of tumor nodules within the prostate to be very limited, especially when different MRI modalities were combined. In the future, randomized trials will demonstrate if delivering an extra boost dose to the gross tumor volume within the prostate results into an improved disease-free survival in patients who are undergoing radiation treatment. In conclusion, with this dissertation we investigated different methods for early detection of nodal involvement in newly diagnosed prostate cancer patients. We conclude that surgical staging is still the most reliable approach at this time and that an extensive lymphadenectomy is warranted to maximize the chances for correct nodal staging. Secondly, we demonstrated that functional imaging can be addressed to detect and localize macroscopic tumor nodules within the prostate. When making use of multiparametric MRI for this purpose, the additional value of choline PET-CT is very limited.