Tumor and Ablation Margin Visibility during Cryoablation of Musculoskeletal Tumors: Comparing Intraprocedural PET/CT Images with CT-Only Images

Abstract

Purpose:

To compare tumor and ice-ball margin visibility on intraprocedural positron emission tomography (PET)/ computed tomography (CT) and CT-only images and report technical success, local tumor progression, and adverse event rates for PET/CT-guided cryoablation procedures for musculoskeletal tumors.

Materials and Methods:

This Health Insurance Portability and Accountability Act (HIPAA)–compliant and institutional review board–approved retrospective study evaluated 20 PET/CT-guided cryoablation procedures performed with palliative and/or curative intent to treat 15 musculoskeletal tumors in 15 patients from 2012 to 2021. Cryoablation was performed using general anesthesia and PET/CT guidance. Procedural images were reviewed to determine the following: (a) whether the tumor borders could be fully assessed on PET/CT or CT-only images; and (b) whether tumor ice-ball margins could be fully assessed on PET/CT or CT-only images. The ability to visualize tumor borders and ice-ball margins on PET/CT images was compared with that on CT-only images.

Results:

Tumor borders were fully assessable for 100% (20 of 20; 95% CI, 0.83–1) of procedures on PET/CT versus 20% (4 of 20; 95 CI, 0.057–0.44) of procedures on CT only (P < .001). The tumor ice-ball margin was fully assessable in 80% (16 of 20; 95% CI, 0.56–0.94) of procedures using PET/CT versus 5% (1 of 20; 95% CI, 0.0013–0.25) of procedures using CT only (P < .001). Primary technical success was achieved in 75% (15 of 20; 95% CI, 0.51–0.91) of procedures. There was local tumor progression in 23% (3/13; 95% CI, 0.050–0.54) of the treated tumors with at least 6 months of follow-up. There were 3 adverse events (1 Grade 3, 1 Grade 2, and 1 Grade 1).

Conclusions:

PET/CT-guided cryoablation of musculoskeletal tumors can provide superior intraprocedural visualization of the tumor and ice-ball margins compared with that provided by CT alone. Further studies are warranted to confirm the long-term efficacy and safety of this approach.

Percutaneous image-guided ablation to treat musculoskeletal (MSK) tumors is increasingly used for pain palliation (1,2) to control oligometastatic disease (3–5) and other indications (6,7). Cryoablation has emerged as a particularly promising ablation modality, with a major advantage being that the ice ball can be visualized and monitored in most tissues. This enables more precise ablation zone control and can reduce the risk of injury to adjacent nerves and other critical structures (8).

However, it often remains difficult or impossible to confirm adequate tumor margins intraprocedurally even with cryoablation. The target tumor itself can be difficult to visualize using standard computed tomography (CT) or ultrasound image guidance. In the bone, there may be no sclerotic or lytic CT correlate for metabolically active tumors initially detected by positron emission tomography (PET)/CT or magnetic resonance (MR) imaging. Even in cases with a CT correlate, the extent of sclerosis or the lytic component may overestimate or underestimate the true extent of viable tumor (9). In soft tissues, the tumor is often obscured by adjacent muscle or other similarly attenuating tissues. In addition, ablation-induced ice, bleeding, and/or edema can further diminish tumor visibility intraprocedurally and preclude margin evaluation. This can lead to incomplete tumor coverage, which is especially problematic in the oligometastatic disease control paradigm in which there is curative intent.

Intraprocedural F-18 fluorodeoxyglucose (FDG) PET/ CT could help overcome this limitation. Recent studies (10,11) have demonstrated improved intraprocedural tumor visualization and margin assessment during cryoablation of tumors in the liver and other sites. Given that PET/CT offers excellent visualization of MSK metastases for many tumor types, it follows that PET/CT guidance for MSK tumor ablation could be similarly beneficial.

The purpose of this study was to compare tumor and ablation margin visibility on intraprocedural PET/CT and CT-only images and report technical success, local tumor progression, and adverse event rates for PET/CT-guided cryoablation procedures for MSK tumors.

MATERIALS AND METHODS

Study Cohort

This was a Health Insurance Portability and Accountability Act (HIPAA)–compliant and Brigham and Women’s Hospital (BWH) institutional review board–approved retrospective, observational, comparative, longitudinal study. Retrospective review was performed of the ablation procedure database of the authors’ institution containing data on all percutaneous ablation procedures performed at the institution between 2012 and 2021. The inclusion criterion was MSK tumor cryoablation procedures performed under PET/CT guidance; noncryoablation procedures (radiofrequency ablation, irreversible electroporation, and microwave ablation) and procedures using other image guidance modalities (MR imaging and CT only without PET) were excluded (Fig 1). This identified 15 patients (mean age, 67 years; age range, 44–97 years; 9 women and 6 men) who underwent 20 PET/CT-guided cryoablation procedures to treat 15 MSK tumors at the authors’ institution (Fig 1, Table 1). Each procedure treated only 1 tumor. Three tumors in 3 patients were treated more than once owing to local progression.

Figure 1.

Derivation of positron emission tomography (PET)/ computed tomography (CT)–guided musculoskeletal (MSK) tumor cryoablation procedures performed from 2012 to 2021. MR = magnetic resonance.

Table 1.

Characteristics of 15 Patients with 15 Tumors Treated I with 20 Procedures (Positron Emission Tomography/Computed I Tomography-Guided Cryoablation of Musculoskeletal Tumors)

Characteristic	Value
Patient characteristics (n = 15)
Age (y)
Median	67
Range	44–97
Sex
Female	9
Male	6
Tumor characteristics (n = 15)
Tumor histology
Lung	5
Ovarian	5
Endometrial	1
Bladder	1
Testicular	1
Renal cell	1
Colon	1
Tumor location
Bone	9
Sternum/Retrosternum	2
Iliac	2
Scapula	1
Clavicle	1
Sacrum	1
Rib	1
Femur	1
Soft tissue	6
Abdominopelvic wall	5
Perispinal	1

All patients were evaluated in the authors’ interventional radiology clinic prior to their procedures, and written informed consent was obtained. PET/CT guidance was selected in cases in which the tumors were expected to be better visualized on intraprocedural PET/CT than on CT or ultrasound alone on the basis of a review of prior imaging and/or tumor type. None of the patients had contraindications to percutaneous ablation, such as an Eastern Cooperative Oncology Group performance status of >2, uncorrectable coagulopathy, acutely decompensated cardiovascular or pulmonary illness, acute renal insufficiency, or pregnancy.

Procedural and Imaging Technique

All procedures were performed using general anesthesia and a PET/CT scanner (Biograph mCT40; Siemens Healthineers, Erlangen, Germany) equipped with single-rotation CT fluoroscopy. A single-bed–position intraprocedural PET/CT was obtained with 1–2-minute PET acquisitions and 5-mm slice thickness. PET in-plane transverse spatial resolution was 4.5 mm (at 1 cm from isocenter), 5.2 mm (10 cm from isocenter), and 6.1 mm (20 cm from isocenter). Intraprocedural CT acquisitions for PET/ CT were obtained using diagnostic quality helical CT Care Dose 4D (Siemens, quality reference: 100 mAs, 120 kVp), 5-mm slice thickness, and 0.5-second rotation time. CT inplane resolution was 16.4 line pairs/cm. All PET/CT acquisitions were performed under expiratory breath-hold (suspended ventilation). Single-rotation CT fluoroscopy was acquired using default 50 mAs, 120 kVp, 4.8-mm slice thickness, and 3 slices per acquisition. No duplicate/ unnecessary acquisitions were obtained specifically for this study.

FDG (mean administered activity, 5.59 mCi; range, 4.4– 10 mCi) was administered intravenously as a single dose approximately 1 hour prior to each procedure. A planning single-bed–position PET/CT was obtained to identify the target lesion. Subsequently, 2–5 cryoprobes (Boston Scientific, Marlborough, Massachusetts) were placed to optimize tumor coverage. Intermittent single-rotation CT-fluoroscopy, CT-only, and/or PET/CT acquisitions were obtained during targeting as needed. After appropriate cryoprobe placement, the ice ball was monitored using intermittent CT and PET/CT acquisitions. Two freeze-thaw cycles were performed in 16 cases, and 3 freeze-thaw cycles were performed in 4 cases (the third freeze cycles were added when repositioning or adding of probes was required for complete tumor coverage).

For osseous lesions requiring cortical transgression, a bone access kit (Bonopty; AprioMed, Uppsala, Sweden; or Arrow Oncontrol; Teleflex, Wayne, Pennsylvania) was initially used, and cryoprobes were subsequently placed coaxially. In 2 cases, introducer needles from the bone access kit were exchanged over a wire for peel-away sheaths, through which the cryoprobes were then placed. Protective strategies, including nerve monitoring (n = 2 cases), hydrodissection (n = 11 cases), warm gauze application for skin warming, and/or needle torquing, were used to protect adjacent critical structures. An intentional pneumothorax was used in 1 case to access a sternal/retrosternal lesion from a lateral approach.

Patients were typically discharged on the same day after several hours of observation or after overnight observation. They were subsequently followed in the authors’ interventional radiology clinic, with postprocedural imaging obtained on a per-patient basis.

PET/CT versus CT-Only Performance Comparison

Procedural images were reviewed independently by 2 interventional radiologists (E.A.B., S.K.B.) to determine the following binary assessments: (a) whether the tumor and its borders could be fully assessed on PET/CT and/or CT-only images for each procedure (Fig 2) and (b) whether the tumor ice-ball margin could be fully assessed on PET/CT and/or CT-only images for each procedure (Fig 3). The tumor ice-ball margin was characterized as fully assessable if the extent of both the tumor and final ice ball could be seen during the procedure.

Figure 2.

Comparison of computed tomography (CT)–only images with positron emission tomography (PET) and fused PET/CT images to compare tumor visibility. Case 1: ovarian cancer metastasis in the rectus muscle (arrows)—the medial margin was obscured by the adjacent rectus muscle (asterisks) on CT but well delineated on PET/CT. Case 2: lung cancer metastasis in the sacrum (white arrows)—margins were underestimated by the lytic lesion on CT but were better seen to extend medially and posteriorly (asterisks) on the PET/CT. Case 3: bladder cancer metastasis in the left iliac bone (arrows)—the lesion demonstrated a mixed lytic and sclerotic appearance on CT, but PET/CT confirmed only the lytic component to be metabolically active tumor. The sclerotic portion (asterisks) was not metabolically active.

Figure 3.

Comparison of intraprocedural computed tomography (CT)–only images with positron emission tomography (PET)/ CT images during cryoablation. Case 1: testicular cancer metastasis in the right paraspinal region—CT-only images demonstrated the ice ball (arrows), but the tumor was not seen. On the fused PET/CT image, the tumor (asterisk) was well seen and the ice-ball margin was assessable circumferentially by examining how far the ice ball (arrows) extended beyond the tumor. Case 2: renal cell cancer metastasis in the scapula—the CT-only image showed the ice ball (arrows); however, the isodense metastasis was obscured within the ice ball. The PET/CT image showed the tumor (asterisk) engulfed by the ice ball (arrows). Again, the ablation margin could be inferred by how far the ice extended beyond the tumor over the entire tumor circumference.

For each procedure, images were reviewed independently on a picture archiving and communication system (PACS) workstation (Visage Imaging, San Diego, California). For PET/CT assessment, 3-mm axial images of the unfused attenuation-corrected PET and CT images as well as fused overlaid PET/CT images were analyzed to determine whether the tumor and ice-ball margin could be fully assessed. For CT-only assessment, the diagnostic quality CT-only images were evaluated using 3-mm axial slices, with 3-dimensional multiplanar reconstruction (postprocessed using Visage software) also used as needed. For CT-only images, tumors were defined to be fully visible when a perceptible difference in density (Hounsfield units) between the tumor and surrounding tissues could be observed circumferentially. For PET/CT images, tumors were defined to be fully visible when a perceptible difference in PET avidity between the tumor and surrounding tissues could be observed circumferentially. Ice-ball margins were defined to be fully visible when tumors were fully visible as defined earlier, and there was a perceptible density difference between the ice ball and surrounding unablated tissues. Tumor and ice-ball margin visibility were compared as described further in the statistics section.

Procedural Outcome Analysis

Primary technical success was defined per procedure as complete ice-ball coverage of the entire tumor with at least a 5-mm margin on intraprocedural imaging (PET/CT) or no evidence of residual tumoral enhancement on initial postprocedural imaging (contrast-enhanced MR imaging or CT) (12,13). On intraprocedural imaging, the ice-ball margin was measured from the outer edge of the FDG-avid tumor to the outer extent of the hypodense ice ball on the fused PET/CT images. Ice-ball coverage of the tumor with a 5-mm margin was used as a surrogate measure of ablation zone tumor coverage because the lethal ablation zone is estimated to be within 5 mm of the ice ball (8,14). On postprocedural imaging, residual tumoral enhancement was defined as nodular enhancement corresponding to the site of the original tumor based on manual coregistration. The primary technical success rate was calculated and reported as a percentage of total procedures.

Local progression was defined per patient/tumor as any evidence of progressive disease in or adjacent to the ablation zone on follow-up imaging (eg, suspicious nodular enhancement or new growth). Treated tumors with <6 months of imaging follow-up were excluded from this analysis unless local progression was detected within 6 months. The local progression rate was calculated as a percentage of treated tumors that developed local progression. Adverse procedural events were determined by review of the electronic medical record.

Statistics

The primary analysis tested the effect of image guidance modality—a dichotomous independent variable (PET/CT or CT)—on 2 dichotomous dependent variables: full visibility of the tumor margin (yes or no) and full visibility of the ablation margin (yes or no). The percentage of procedures in which the tumor and ablation margin were fully visible was calculated for PET/CT and CT only, reported as continuous variables, and compared using the exact McNemar test, with statistical significance defined as a P value of <.05. Interreader agreement was quantified by calculating observation agreement percentage and Conger κ index.

Secondary analyses included primary technical success reported as a ratio and percentage of total procedures as a continuous variable, Kaplan-Meier 6-, 12-, and 24-month local tumor progression–free survival with 95% CIs, local progression rate reported as a ratio and percentage of patients as a continuous variable, and adverse events reported descriptively and as ordinal variables on the basis of the Society of Interventional Radiology classification (15).

All statistical analyses were performed using R 4.2.0 (16). The exact McNemar test was performed using the R package “exact2×2” (17), agreement percentage and Conger κ index were calculated using the R package “irrCAC” (18), and survival analysis was performed using the R package “survival” (19).

RESULTS

PET/CT versus CT-Only Performance Comparison

Table 2 summarizes results from the PET/CT versus CT comparison of tumor and ablation margin visibility. Tumors and their borders were fully assessable during 100% (20 of 20; 95% CI, 0.83–1) of the procedures using fused intraprocedural PET/CT compared with 20% (4 of 20; 95% CI, 0.057–0.44) of the procedures using CT-only (P < .001) images. Tumor borders were not well seen by CT owing to isodensity to adjacent muscle or other soft tissues (n = 13) or poor correlation between sclerotic/lytic features (CT) and metabolically active osseous tumor (PET) (n = 3) (Fig 2).

Table 2.

Comparison of Tumor and Ice-Ball Visibility on Computed Tomography versus Positron Emission Tomography/Computed Tomography Images

Imaging metric	Ratio of cases, n/N (%)	Interreader agreement (%); κ
Tumor border visibility
PET/CT	20/20 (100)	100; n/a
CT only	4/20 (20)	95; 0.83
P value	<.001
Ice-ball visibility
PET/CT	16/20 (80)	90; 0.69
CT only	1/20 (5)	100; 1
P value	<.001

CT = computed tomography; n/a = not applicable; PET = positron emission tomography.

Tumor ablation margins were fully assessable for 80% (16 of 20; 95% CI, 0.56–0.94) of the procedures using PET/ CT compared with 5% (1 of 20; 95% CI, 0.0013–0.25) of the procedures using CT-only (P< .001) images (Fig 3). In all 4 cases, the ablation margin could not be assessed by PET/CT because of poor ice visibility in the bone. The ablation margin was not assessable on CT-only images owing to several factors, such as the aforementioned initial inability to visualize the tumor extent (n = 15), loss of tumor visibility after start of ablation (n = 2), and/or inability to visualize the ice ball (n = 4).

Interreader agreement was 95% (Conger κ index, 0.83), 100%, 100% (Conger κ index, 1.0), and 90% (Conger κ index, 0.69) for CT-only tumor visibility, PET/CT tumor visibility, CT-only ablation margin visibility, and PET/CT ablation margin visibility, respectively.

Procedural Outcome Analysis

Primary technical success was achieved in 75% (15 of 20; 95% CI, 0.51–0.91) of the procedures in which intraprocedural and/or initial follow-up imaging was sufficient to confirm adequate ablation coverage. In 5% (1 of 20; 95% CI, 0.0021–0.25) of the procedures, complete ablation was not achieved owing to the inability to displace the bowel adjacent to an abdominal wall mass despite hydrodissection. In 20% (4 of 20; 95% CI, 0.057–0.43) of the procedures, primary technical success could not be confirmed, although no subsequent local progression occurred. In these 4 cases, the intraprocedural ice ball was not well seen in the bone. In addition, initial postprocedural imaging (at least 1 day later) was performed with PET/CT rather than contrast-enhanced CT or MR imaging, and complete tumor coverage could not be confirmed owing to confounding FDG avidity from periprocedural inflammation.

For the 15 patients treated with ablation for 15 tumors, the 6-month, 12-month, and 24-month local tumor progression–free survival rate was 82.5% (63.1%–100%), 70.7% (47.2%–100%), and 70.7% (47.2%–100%), respectively (Table 3). Thirteen patients had at least 6 months of follow-up after the initial treatment. Of these 13 patients, local tumor progression was observed in 23% (3 of 13; 95% CI, 0.050–0.54). One patient with aggressive non–small cell lung cancer involving the right lateral abdominal wall locally showed progression 4 times. Each local progression was treated with repeat ablation, with no evidence of local progression after the final fifth ablation procedure. Another patient with ovarian cancer involving the left anterior abdominal wall/rectus musculature had local progression twice. Each occurrence was treated with repeat ablation, with no evidence of local progression after the final procedure. A third patient with non–small cell lung cancer involving the sacrum near the S3 neuroforamina underwent ablation for pain control. The patient’s pain initially improved, but his tumor locally progressed at 7 months. Owing to proximity to adjacent nerves, it was not amenable to repeat ablation. Therefore, accounting for repeat ablation, secondary treatment efficacy was 92% (12 of 13, 95% CI, 0.64–1.00)—12 tumors in 12 patients demonstrated durable local control with ablation, whereas 1 tumor in 1 patient had local progression not amenable to further ablation.

Table 3.

Details and Outcomes of 20 Positron Emission Tomography/Computed Tomography-Guided Ablation Procedures in 15 Patients

Procedural details (n = 20)	Value
Indication
Curative only	11
Palliative only	3
Both	6
Protective maneuvers
Hydrodissection	11
Nerve monitoring	2
Intentional pneumothorax	1
Procedural outcomes
Primary technical success	15/20 (75)
Adverse events	3/20 (15)
Grade 1	1/20 (5)
Grade 2	1/20 (5)
Grade 3	1/20 (5)
Patient/tumor outcomes
Treatment efficacy (at least 6 mo of follow-up)
Primary	10/13 (77)
Secondary	12/13 (92)
KM local progression-free survival
6 mo	83 (63–100)
12 mo	71 (47–100)
24 mo	71 (47–100)

Note-Values are reported as n, n/N (%), or % (95% CI). KM = Kaplan-Meier.

Three adverse events occurred, including 1 Grade 3 (readmission for pain control), 1 Grade 2 (small non-displaced iliac wing fracture requiring crutches), and 1 Grade 1 (reactive self-limited pleural effusion after chest wall ablation). There were no Grade 4 or 5 adverse events.

DISCUSSION

In this study, PET/CT guidance significantly improved visualization of the tumor borders and ablation margins intraprocedurally compared with standard CT-only imaging. PET/CT proved to be useful for a variety of FDG-avid tumor types in a variety of locations in the bone and soft tissues. Tumors were fully visible with CT alone in only 21% of the procedures because they were frequently obscured by adjacent normal tissues with similar attenuation, did not have a clear sclerotic/lytic correlate, had a correlate that was not fully representative of the true tumor extent, or were obscured by treatment-related changes (eg, bleeding, ice, and inflammation) during the procedure. In comparison, tumors were well seen during all procedures with PET/CT, which allowed precise targeting and needle positioning. There was also persistent tumoral FDG avidity during cryoablation, consistent with findings from prior reports (10,11,20) in the liver and lung. This enabled the tumor extent and ice-ball margin to be assessed during cryoablation.

The visible ice-ball edge represents the 0° isotherm. The true lethal zone is dependent on various factors, including tissue type, vascular perfusion, and freeze-thaw cycle protocol, but has been generally reported to be approximately 5 mm inside of the visible ice-ball perimeter (8,14). Therefore, ice-ball coverage of the tumor with at least a 5-mm margin served as a surrogate measure of adequate ablation coverage. The ablation zone coverage of tumor could not be fully assessed using this approach in only 4 procedures with PET/CT. In all 4 cases, this was due to poor visibility of the ice in the bone. Despite this limitation, excellent visualization of the tumor borders allowed the probes to be placed with confidence under PET/CT guidance, and there was evidence of local progression in only 1 of these 4 cases.

Overall, 2 patients with apparent confirmation of adequate intraprocedural ablation margins had multiple local progressions, 1 with aggressive non–small cell cancer and another with ovarian cancer. In these cases, there may have been microscopic tumor foci that were below the detection threshold of PET/CT. This has been previously reported in the liver (21) and suggested to be analogous to R1 surgical resection with microscopic residual disease (22). In the third patient who had local progression, ablation was performed for pain palliation, and the ice-ball margin could not be assessed owing to poor visibility of ice in the bone.

The 92% secondary treatment efficacy rate seen in this study is noted relative to prior studies without PET guidance showing rates of 87% (5) and 86% (23). The small and heterogeneous cohorts preclude definitive conclusions about the overall achievable control with PET/CT versus CT guidance; a prospective comparative study may be warranted. Only 3 adverse events were observed, all being less than Grade 4. This is comparable with prior reports demonstrating the safety of cryoablation for MSK tumors (23,24). Improved tumoral visibility also enabled more precise probe placement to minimize nontarget ablation zone extension; this also likely contributed to the overall low adverse event rate observed in this study.

There were several limitations. This was a retrospective feasibility study with a small sample size and relatively short follow-up interval. Larger studies with a longer follow-up period would be helpful to confirm long-term outcomes, efficacy, and safety. This study did not compare local tumor recurrence rates for PET/CT with those for CT-only–guided MSK tumor ablation, and therefore, the relative clinical benefit of PET/CT-guided ablation for local tumor recurrence is not yet delineated. The choice of PET/ CT as the guidance modality was operator-dependent and necessarily a subjective decision in this retrospective study. The authors expect that PET/CT may not be as useful in tumors with low metabolic activity or extensive necrosis. Although 2 reviewers (E.A.B., S.K.B.) independently reviewed all images with good interreader agreement, observer bias may be possible given the small sample size and nonblinded image review. Finally, because most patients were treated for oligometastatic disease control, pain control outcomes were not quantitatively reported. Pain relief after cryoablation has been well established in larger series, and the ablation methods used in this study were similar (1,24).

In conclusion, PET/CT-guided cryoablation of MSK tumors was found to provide superior intraprocedural visualization of the target lesion and assessment of the tumor ablation margin compared with CT-guided cryoablation. Further studies are warranted to confirm the long-term efficacy and safety of this approach.

RESEARCH HIGHLIGHTS.

• Positron emission tomography (PET)/ computed tomography (CT)–guided cryoablation was used to treat 15 musculoskeletal tumors in 15 patients.

• PET/CT was compared with CT alone in delineating tumor and ice-ball margins.

• PET/CT provided superior visualization of the target tumor borders compared with CT-only images (fully assessable margins in 100% vs 20% of procedures; P < .001).

• PET/CT provided superior assessment of the ablation margin compared with CT-only images (fully assessable margins in 80% vs 5% of procedures; P < .001).

STUDY DETAILS.

Study type: Retrospective, observational, descriptive study

Level of evidence: 4 (SIR-D)

ABBREVIATIONS

CT

computed tomography

FDG

F-18 fluorodeoxyglucose

HIPAA

Health Insurance Portability and Accountability Act

MR

magnetic resonance

MSK

musculoskeletal

PACS

picture archiving and communication system

PET

positron emission tomography