Initial Experience with Volumetric ⁶⁸Ga-PSMA I&T PET/CT for Assessment of Whole-Body Tumor Burden as a Quantitative Imaging Biomarker in Patients with Prostate Cancer

DISCUSSION

In the present study, we demonstrated that PSMA-TV and TL-PSMA may be used as quantitative imaging biomarkers in patients with prostate cancer and that these parameters may be applied for therapy monitoring. The use of volumetric parameters in ¹⁸F-FDG PET has gained interest to predict treatment response and overall prognosis in oncologic patients. For instance, van Rossum et al. demonstrated that the change in TLG after induction chemotherapy was predictive for a poor pathologic response in patients with esophageal adenocarcinoma (15). In another study, baseline MTV before chemotherapy for diffuse large B cell lymphoma was the only statistically significant predictor of progression-free survival at multivariate analysis (16). In prostate cancer, volumetric parameters have been evaluated in a small number of studies using radiotracers other than PSMA ligands (22,23). Incerti et al. assessed the value of ¹¹C-choline PET/CT in the context of radiation treatment of lymph node recurrence from prostate cancer and reported that MTV was significantly associated with the clinical relapse-free survival and biochemical relapse-free survival (22).

We have applied the concept of whole-body MTV or TLG assessment into the setting of PSMA imaging of prostate cancer. When the relationship between serum PSA levels and volumetric PET parameters was analyzed, a statistically significant correlation was observed with both whole-body PSMA-TV (r = 0.48, P < 0.0001) and whole-body TL-PSMA (r = 0.52, P < 0.0001), demonstrating that the volumetric parameters were consistent with established biomarkers of metastatic tumor load. All patients had undergone radical prostatectomy before PET/CT-based restaging, and special attention was paid to exclude all patients with residual prostate tissue (e.g., after local cryotherapy), which might have influenced the correlation analysis between volumetric parameters and PSA. Previous studies found that high-grade tumors may produce less PSA per gram of tumor (24), and paradoxically low PSA levels in the context of poor differentiation may be observed, limiting the usefulness of PSA measurements (12). Volumetric PET parameters demonstrated a highly significant correlation with PSA (P < 0.0001), which was driven by Gleason 6 tumors (PSMA-TV, r = 0.98, P < 0.0001) and Gleason ≥ 9 tumors (PSMA-TV, r = 0.53, P = 0.006). By contrast, no significant correlation was found between PSA levels and TV in Gleason 7 and Gleason 8 tumors. There is currently no model to fully explain this phenomenon. Other PET-derived parameters such as SUVs are not useful for assessment of tumor burden, because they characterize only the intensity of tracer uptake in individual lesions. Accordingly, neither SUV_max (r = 0.12, P = 0.22) nor SUV_mean (r = 0.8, P = 0.45) was associated with PSA levels in this study. Recently, novel biomarkers have been investigated for improved detection of prostate cancer. It would be interesting to explore the correlation of metastatic load on PET/CT and these biomarkers, for example, kallikreins (25).

Whole-body PSMA-TV was not significantly associated with primary Gleason scores (P = 0.32), as was to be expected. The size of metastases depends on other factors such as proliferation rate or previous treatment history. By contrast, whole-body TL-PSMA was independently associated with the initial Gleason score (r_partial = 0.33, P < 0.001). Besides volume, TL-PSMA also considers the degree of PSMA expression. In previous studies, high-level PSMA expression has been found in higher Gleason scores (26), providing an explanation for the correlation between Gleason score and TL-PSMA. However, the major determinant of TL-PSMA is also the patient’s individual course of the disease.

Considering potential applications of volumetric PSMA PET imaging, the prediction of therapy response of internal-emitter therapy may be an interesting option. The metastatic cure probability has been demonstrated to depend on the tumor size and the intratumoral activity concentration in radionuclide therapy influencing the absorbed dose (27). TL-PSMA may be an ideal parameter to account for both lesion size and intralesional PSMA expression when predicting metastatic cure. Moreover, treatment planning based on the extent of tumor burden has recently gained prominence in metastatic prostate cancer therapy (28,29). Several clinical trials have used tumor burden for stratification of patients into low- or high-volume disease. In a recent study that assessed whether ADT plus docetaxel would result in longer overall survival than ADT alone, the median overall survival was 17.0 mo longer in the combination group than in the ADT-alone group (49.2 vs. 32.2 mo) in the subgroup of patients with high-volume disease (28). In the absence of a sensitive imaging–derived biomarker, stratification was performed using suboptimal methods, for example, high volume was defined as the presence of visceral metastases or 4 bone lesions or more with 1 or more beyond the vertebral bodies and pelvis in the aforementioned study (28). In the Southwest Oncology Group trials, minimal disease was defined as disease confined to the spine, pelvic bones, or lymph nodes (29). These definitions can provide only an imprecise estimate of whole-body tumor burden. A more sophisticated approach using PSMA-TV or TL-PSMA may considerably refine stratification of patients, providing real volumetric parameters and a more accurate assessment of the tumor burden.

Assessment of volumetric parameters was time- and labor-intensive in this cohort with metastatic prostate cancer, and we therefore restricted the analysis to patients with 10 or fewer metastases. Although our approach consisting of manual placement of volumes of interest may also be used in patients with more than 10 lesions, other technical approaches may facilitate the assessment of volumetric parameters in patients with extensive disease, for example, fully automated, SUV threshold–based whole-body segmentation techniques that allow for a rapid definition of whole-body tumor burden (14).

Some limitations of the present study should be acknowledged. First, the study was retrospective in nature. Second, we evaluated a single threshold for calculation of PSMA-TV and TL-PSMA. It is important to keep in mind that volumetric data generated by PET represent an estimate of the true volume. Although we experimentally determined that a threshold of 45% provided the best results using our imaging and reconstruction protocol, and have used this threshold in previous studies (9), further refinement using different cutoff values, for example, 46% or 47%, may be beneficial. However, the observed correlations are robust and consistent with the literature. Moreover, the lesions chosen for the threshold assessment were representative of the total study population. The precise threshold used in this study may not be generalized to all types of acquisition protocols or reconstruction techniques but provides a proof of concept for volumetric assessment of prostate cancer lesions. Furthermore, ADT may alter PSMA expression by tumor cells and may potentially influence volumetric assessment based on an SUV_max threshold. Third, it would be desirable to obtain additional histopathologic specimens for assessment of the differentiation of metastatic lesions. After previous treatment, the initial Gleason score may not fully reflect the differentiation of metastases. However, this was not possible for both practical and ethical reasons. Finally, we demonstrated the feasibility of volumetric PSMA ligand PET and explored the relation of PSMA-TV and TL-PSMA with serum PSA levels. Future prospective studies will investigate the prognostic impact of these parameters in prostate cancer.