The current prospective study demonstrates that post-neoadjuvant treatment FDG-PET evaluations in adult patients with bone sarcomas identified histopathological treatment responders accurately when a prospective cut-off of ≥60% reductions in SUVmax was applied.
The inability of changes in tumor size to distinguish treatment responders from nonresponders is not surprising and has been well established in bone sarcomas and various types of cancers [16
]. Parameters assessed by conventional radiography such as tumor location, tumor size, radiographic appearance, margination, cortical destruction, and periosteal reaction are of limited use in assessing therapeutic responses [17
As a prerequisite for monitoring of treatment effects most bone sarcomas exhibit markedly increased FDG-uptake with a mean baseline SUVmax of 8.1 ± 4.1
g/mL (range, 3.6 to 15.7
g/mL). This observation is in concordance with two previsously published papers that describe mean SUVs of 9.6 and 7.9, respectively [3
]. Thus, it should be feasible to determine changes in FDG uptake in response to treatment and to use this information for predicting treatment responses. This has been reported in some earlier prospective and retrospective studies [7
]. Schulte et al. [7
] observed in 27 patients with osteosarcoma that a decrease in FDG SUV tumor to background ratio from baseline to end of treatment by at least 60% detected histopathologic response with a sensitivity of 100% and a specificity of 80%.
In the current study we prospectively applied this previously reported threshold as a response criterion. A 60% threshold for a metabolic response correctly classified 7 of 8 histopathologic nonresponders as metabolic nonresponder (specificity, 88%) and 3 of 4 histopathological responders as metabolic responders (sensitivity, 75%). The single “false negative” response by PET occurred in the patient with the lowest baseline tumor FDG uptake of all study participants (SUVmax at baseline: 3.6
g/mL). In this patient SUVmax decreased from baseline to late followup by only 38%, to 2.3
g/mL. This false negative result is consistent with the notion that response assessments by FDG-PET are less reliable in tumors with lower FDG baseline uptake [23
In a subgroup analysis that excluded patients with chondrosarcoma and giant cell tumor the sensitivity of FDG-PET in assessing histopathologic response was identical (sensitivity, 75%). However, excluding patients with those subtypes resulted in a decreased specificity (80% versus 88% resp.).
In a retrospective analysis of 36 patients with Ewing sarcoma family of tumors, Hawkins et al. [18
] found that a decrease in SUV ≥50% and an end of treatment SUVmax of <2.5
g/mL detected histopathologic response with a sensitivity of 83% and 76%, respectively. Despite the superiority of changes in SUV in the detection of histopathologic responses, an end of treatment SUV of less than 2.5
g/mL was more predictive of progression free survival in their study. We therefore tested whether an end of treatment tumor SUVmax of <2.5 could also identify treatment responders. This resulted in a sensitivity of 75% and a specificity of 88% for predicting tissue responses. However, the combination of both response criteria (response: ≥60% decrease in SUVmax or SUVmax at late followup <2.5; nonresponse: <60% decreases in SUVmax and SUVmax at late followup >2.5) correctly identified all histopathological responder as metabolic responder (sensitivity, 100%) and 6 of 8 histopathologic nonresponder as metabolic nonresponder (specificity, 75%).
End of treatment evaluations has limited impact on patient management. We have previously shown that reductions in SUVmax by ≥35% after the initial cycle of neoadjuvant treatment predicts histopathological responses in soft tissue [5
] sarcomas. In another prospective study at our institution such early response assessments were obtained in 4 patients with bone sarcomas. A metabolic response of 35% correctly classified 1 of 1 histopathologic responder as metabolic responder and 2 of 3 histopathologic nonresponders as metabolic nonresponders (). Future studies need to determine whether such early response predictions hold up in larger patient populations.
FDG-PET/CT at baseline, early followup, and after completion of neoadjuvant treatments in a histopathological responder (a) and a nonresponder (b).
PET data can be analyzed semiquantitatively. This allows for a reliable definition of metabolic tumor activity before, during, and after neoadjuvant therapy. However, few studies have attempted to apply response criteria across institutions and study populations. The current study adds to a growing body of evidence established in sarcoma and other cancers that late reductions in SUV by ≥60% can accurately predict responses to neoadjuvant therapy [7
Besides the evaluation of response/nonresponse by PET, new imaging modalities such as dynamic magnet resonance imaging [26
] and new image analysis methods/response criteria [27
] are currently under investigation for early response assessments in sarcoma patients. Previously published data suggest that a combination of changes in tumor density (>15% change) and in modified tumor size (>10%) accurately detects early response in gastrointestinal stromal tumors (GIST) to Imatinib treatment [27
]. Whether these imaging modalities or response criteria are superior over PET and the proposed definition of a metabolic response needs to be addressed in future studies.
The primary limitation of this study is the small study population. However, we felt that reporting this trial adds to growing body of literature that addresses the ability of FDG-PET/(CT) for monitoring treatment responses in sarcoma patients. Larger prospective studies are needed to confirm these data. Another limitation was the heterogeneity of the study population with regards to tumor histology and treatment regimens. However, we believe that any useful imaging test used for monitoring therapeutic interventions needs to be applied across disease subtypes and treatment approaches.
In conclusion, serial FDG-PET imaging accurately classified histopathologic response/nonresponse to neoadjuvant therapy in adult primary bone sarcomas. It is noteworthy that a combination of previously applied response criteria (reductions in SUVmax by ≥60% and a posttreatment tumor SUVmax of <2.5) provided the best response predictions. This information could now be applied prospectively to examine whether PET-guided therapeutic decisions affect survival of bone sarcoma patients.