Fusion of Metabolic Function and Morphology: Sequential [¹⁸F]Fluorodeoxyglucose Positron-Emission Tomography/Computed Tomography Studies Yield New Insights Into the Natural History of Bone Metastases in Breast Cancer

Yong Du,   Journal of Clinical Oncology, Vol 25, No 23 (August 10), 2007: pp. 3440-3447

The contribution of [¹⁸F]fluorodeoxyglucose positron-emission tomography ([¹⁸F]FDG-PET) imaging to the management of a variety of malignancies is now well recognized. However, its role in the evaluation of bone metastases has not been established adequately. Bone metastases are a frequent complication of cancer, occurring in up to 70% of patients with advanced breast cancer, for example. On the basis of radiographic appearance, they are classified as osteolytic, osteoblastic, or mixed-pattern if both elements are presented in the same lesion. The consequences of bone metastasis are often devastating, and novel treatments have been under intensive investigation. Techniques are required to identify patients with active bone metastases and to monitor treatment response in a timely manner.

Bone scan using technetium-99m–labeled diphosphonates has been the method of choice for the detection of bone metastases, albeit with poor specificity. A positive bone scan reflects the existence of osteoblastic response, which may be secondary to either malignant or benign etiology. The osteoblastic response identified on bone scan persists for some considerable time and the scan therefore remains positive even if there is good response to treatment. Bone scan is thus unsuitable to evaluate treatment response. Other imaging modalities, such as computed tomography (CT) and magnetic resonance imaging, have also been found useful in the detection of bone metastases, but there has been no satisfactory technique for monitoring therapy response.The most recent Response Evaluation Criteria in Solid Tumors guidelines still consider bone metastases to be "nonmeasurable."

[¹⁸F]FDG has been established as a PET imaging tracer for the detection and monitoring of numerous malignancies, owing to the increased glycolysis of most tumor cells. However, the results obtained when investigating [¹⁸F]FDG-PET in the detection of bone metastases have been conflicting, with sensitivity varying widely from 56.5% to 100%. Cook  compared the [¹⁸F]FDG-PET findings with x-ray CT and observed a correlation between the [¹⁸F]FDG uptake features and the CT morphologies of bone metastases from breast cancer. [¹⁸F]FDG-PET was found to be superior to bone scan in the detection of osteolytic metastases, but significantly less sensitive in the detection of osteoblastic lesions (P < .05). This observation has been confirmed in subsequent studies. However, most of the patients included in these studies had been treated with systemic therapy, and it has been suggested that osteolytic bone metastases may become osteoblastic after effective treatment.Therefore, an important unanswered question that needs to be addressed is the clinical relevance of [¹⁸F]FDG-negative osteoblastic lesions, which are often positive on bone scan. The recently available hybrid PET/CT enables us to correlate directly the [¹⁸F]FDG uptake features with the CT morphology, and this would ideally be investigated by sequential [¹⁸F]FDG-PET/CT studies performed on the same patients treated during a certain time period.

In this study, by following up retrospectively the same bone metastases with sequential [¹⁸F]FDG-PET/CT studies, we correlated the treatment status with the temporal changes in [¹⁸F]FDG uptake and the corresponding CT morphology of each bone lesion, thereby investigating the clinical relevance of [¹⁸F]FDG uptake features of bone metastases with various radiographic appearances.

Patients and Methods: Bone metastases were found in 67 of 408 consecutive patients with known/suspected recurrent breast cancer on [¹⁸F]FDG-PET/CT, characterized by CT morphology changes and/or bony [¹⁸F]FDG uptake. Twenty-five of the patients had sequential [¹⁸F]FDG-PET/CT examinations (86 studies) over an average follow-up period of 23 months. The temporal changes in [¹⁸F]FDG uptake and corresponding CT morphology features of 146 bone lesions identified in these 25 patients were followed up and correlated with therapeutic outcome retrospectively.

Results: The 146 lesions were classified as osteolytic (77), osteoblastic (41), mixed-pattern (11), or no change/negative (17) on CT. The majority of the osteolytic (93.5%) and mixed-pattern lesions (81.8%), but fewer of the osteoblastic lesions (61%), showed increased [¹⁸F]FDG uptake. After treatment, 58 osteolytic lesions (80.5%) became [¹⁸F]FDG negative and osteoblastic on CT and only 14 relatively large lesions (19.5%) remained [¹⁸F]FDG avid. Of the 25 [¹⁸F]FDG-avid osteoblastic lesions, 13 (52%) became [¹⁸F]FDG negative, but 12 (48%) remained [¹⁸F]FDG avid and increased in size on CT. Five of the mixed-pattern lesions remained [¹⁸F]FDG avid after treatment. All 17 CT-negative lesions became [¹⁸F]FDG negative; however, nine of them became osteoblastic. None of the initially [¹⁸F]FDG-negative lesions showed [¹⁸F]FDG avidity during follow-up.

By monitoring the [¹⁸F]FDG uptake changes of 146 individual bone metastases together with corresponding CT appearances, we have been able to reveal the heterogeneous nature of bone metastases. Of the 25 patients investigated, 18 presented with a mixture of osteolytic, osteoblastic, mixed-pattern, or CT-negative lesions on CT before additional treatment, whereas only six patients presented with osteolytic lesions only. These lesions showed a wide range of [¹⁸F]FDG uptake intensity. After treatment, 76.4% of [¹⁸F]FDG-avid lesions became [¹⁸F]FDG negative; however, most of them remained abnormal on CT, with predominantly osteoblastic appearances, and most also remained positive on bone scan.

Normal bone is subject to continuous remodeling through the coordinated activity of osteoclasts and osteoblasts. The classification of bone metastases as osteolytic or osteoblastic actually represents two extremes of a continuum of dysregulated bone remodeling. It is reported that the majority of patients with breast cancer have predominantly osteolytic lesions, whereas approximately 15% to 20% of them have predominantly osteoblastic lesions. Breast cancer cells are understood to produce cellular factors that directly or indirectly induce the formation of osteoclasts, but suppress the osteoblast activity. In turn, bone resorption by osteoclasts releases growth factors from the bone matrix that stimulate tumor growth. Such a feedback has been observed in animal experiments in which treatment of myeloma in mice with agents that block bone resorption but have no direct effect on tumor growth not only inhibits the formation of osteoclasts but also decreases the tumor burden. By monitoring bone metastases with sequential [¹⁸F]FDG-PET/CT imaging, we have observed that the responding osteolytic or CT-negative lesions gradually became osteoblastic. This suggests the possible existence of a reversal of the above-described feedback: in eradicating tumor cells, effective anticancer therapy shifts the balance to favor osteoblast activity and bone formation.

The mechanisms of osteoblastic metastasis remain unknown. It can be postulated that tumor cells produce factors stimulating the activity of osteoblasts. In this study, two patients who received no previous systemic therapy presented with predominately osteoblastic metastases that were all [¹⁸F]FDG avid, and both patients died with progressive metastases. when osteoblastic metastases progressed, while remaining [¹⁸F]FDG positive, the metastases indeed became more osteoblastic on CT and larger in size. However, the coexistence of progressive osteolytic and osteoblastic metastases in some patients in this study suggests the complexity of the mechanisms regulating the osteoclast/osteoblast balance in bone metastases.

Most previous [¹⁸F]FDG-PET studies have focused mainly on evaluating the sensitivity of [¹⁸F]FDG imaging in detecting bone metastases. In such cases, the [¹⁸F]FDG-PET sensitivity was calculated against other imaging findings, mostly from bone scans, which could not differentiate post-treatment bony changes from active bone metastases. In addition, these studies included patients with varied treatment backgrounds. Without sufficient follow-up data, these studies could not elucidate the relationship between the temporal changes in [¹⁸F]FDG uptake and the heterogeneous radiographic appearances of bone metastases. Therefore, conflicting results have been reported, with [¹⁸F]FDG-PET sensitivity ranging widely. Notably, recent studies have suggested that serial [¹⁸F]FDG-PET is useful in monitoring bone metastasis response to anticancer therapy, but unlike our study, these reports do not include radiographic data of the bone lesions.

It has been reported that osteolytic metastases have significantly higher [¹⁸F]FDG uptake than osteoblastic lesions, with a hypothesis that osteolytic lesions contain higher tumor cellularity. However, in these studies a considerable percentage of the osteoblastic lesions were [¹⁸F]FDG negative, which would inevitably have lowered the mean SUV of osteoblastic lesions. We have demonstrated that [¹⁸F]FDG-negative lesions, most of which are osteoblastic, are likely to represent post-treatment osteoblastic change rather than active tumor. Therefore, when the [¹⁸F]FDG uptake intensity was analyzed in this study, only the [¹⁸F]FDG-avid lesions were included. Although the mean SUV_max of osteolytic lesions seems higher, this difference is not significant. Instead, the SUV_max seems more related to the size of the lesion, with larger lesions more likely to have a higher SUV_max, probably owing to the partial volume effect. However, unlike the study performed by Cook when the SUV_max was measured in this study, partial volume effects were not corrected because this approach is more concordant with clinical practice.

We found that the presence of persistently [¹⁸F]FDG-avid lesions correlated with poorer prognosis. We also found that osteoblastic lesions that were [¹⁸F]FDG avid seem more resistant to treatment. Our results initially seem to conflict with the observations by Cook that patients with osteoblastic or mixed-pattern disease have a better prognosis. As mentioned, however, in the study by Cook et al, all of the osteoblastic lesions were included, and a large percentage of them were not [¹⁸F]FDG avid. Our data suggest that this type of lesion is in fact more likely to represent post-treatment change rather than active tumor, and assessment of persistently [¹⁸F]FDG-avid lesions seems more useful in evaluating the effectiveness of anticancer therapy.

The availability of corresponding high-resolution CT images in hybrid PET/CT improves the sensitivity and specificity of [¹⁸F]FDG-PET imaging, and the interobserver variability in this study has been minimal. With only [¹⁸F]FDG-PET images, it is usually difficult to identify small lesions and localize them accurately to the skeleton because of the limited spatial resolution of the technique. In this study, we were able to identify 38 bone metastases smaller than 20 mm, with the smallest lesion measuring 4.8 mm (maximum long axis). When the corresponding bone scans were reviewed, even with knowledge of the [¹⁸F]FDG-PET/CT findings, 26 of these lesions were still either invisible or nonspecific on bone scan.

However, we have found that the diagnosis of bone metastases showing no CT morphologic change (CT negative) poses a challenge, and there were three preconsensus discordant lesions in this category (one excluded on consensus). It has been reported that such metastases could account for a much higher percentage in certain malignancies, such as lymphoma.This reflects the existence of intrinsic limitations of [¹⁸F]FDG-PET/CT imaging.

In conclusion, 146 bone deposits from 25 patients with breast carcinoma were followed up for a median period of 23 months. Appearances on [¹⁸F]FDG-PET and CT images were investigated and compared. [¹⁸F]FDG-PET/CT hybrid imaging was found to better reflect tumor activity of bone metastases. Prospective trials to confirm and expand our initial observations in bone metastases from breast cancer, and to investigate whether these observations apply to other cancers, are clearly warranted.

Conclusion: [¹⁸F]FDG uptake reflects the immediate tumor activity of bone metastases, whereas the radiographic morphology changes vary greatly with time among patients.