Non-Invasive Imaging Biomarkers to Predict the Hepatopulmonary Shunt Fraction Before Transarterial Radioembolization in Patients with Hepatocellular Carcinoma

Figures & data

Figure 1 Contrast-enhanced axial CT images of the liver in (A–C) display arterial and venous sequences of patients with different growth patterns and imaging biomarkers of hepatocellular carcinoma (HCC): non-rim arterial phase enhancement (arrow), washout with tumor-liver difference (dotted arrow), internal tumor arteries (arrowheads), tumor capsule (*), hypodense halo (†), peri-tumoral enhancement (‡), and index tumor (dotted circle).

Figure 2 Overview of the segmentation process in a 79-year-old patient with multiple cirrhotomimetic hepatocellular carcinoma liver lesions (>30 in total). The five largest lesions in diameter (#; index lesion marked as #1) were segmented to analyze the volume. (A–C) display axial arterial, portal venous, and venous CE-CT sequences without tumor masks, while (D–F) include the generated tumor segmentations (green masks). Finally, (G) shows the coronal view of the liver with tumor masks of three of the largest HCC lesions, while (H) shows the 3D tumor masks of the five largest HCC lesions (abbreviations in (H): R = right, L = left, S = superior, I = inferior).

Table 1 Patient and Disease-Specific Characteristics and Treatment Data

Characteristic	Value
Total number of included patients	56
Age at time of TARE evaluation (years)
Mean ± standard deviation	66.49 ± 11.72
Range	25–85
Sex (n)
Male	46 (82.1%)
Female	10 (17.9%)
Body mass index (kg/m^2)
Mean ± standard deviation	26.72 ± 4.76
Range	15.62–41.1
Barcelona Clinic Liver Cancer stage (n)
A	11 (19.6%)
B	33 (58.9%)
C	4 (7.1%)
D	2 (3.6%)
Not available	6 (10.7%)
Child–Pugh class (n)
A	50 (89.3%)
B	2 (3.6%)
C	0
Not available	4 (7.1%)
Eastern Cooperative Oncology Group performance status (n)
0	43 (76.8%)
1	3 (5.4%)
2	2 (3.6%)
3	1 (1.8%)
4	1 (1.8%)
Not available	6 (10.7%)
α-Fetoprotein (µg/L)
Mean ± standard deviation	11,361 ± 28,069
Range	1–126,055
Total bilirubin (mg/dL)
Mean ± standard deviation	0.83 ± 0.48
Range	0.26–2.2
Serum albumin (g/L)
Mean ± standard deviation	22.9 ± 17.18
Range	3.1–46.1
International normalized ratio
Mean ± standard deviation	1.13 ± 0.16
Range	0.91–1.94
Ascites (n)
Minor	8
Moderate	7
Encephalopathy (n)	0
Underlying disease (n)
Hepatitis B virus	6 (10.7%)
Hepatitis C virus	7 (12.5%)
Hepatitis B and hepatitis C virus	3 (5.4%)
Hepatitis B virus and alcohol	1 (1.8%)
Hepatitis C virus and alcohol	3 (5.4%)
Alcoholic liver disease	12 (21.4%)
Halogenated hydrocarbons	1 (1.8%)
Non-alcoholic steatohepatitis	2 (3.6%)
Not available	21 (37.5%)
Extrahepatic disease (n)	20 (35.7%)
Prior therapies (n)
Systemic
Sorafenib	12 (21.4%)
Tenofovir	2 (3.6%)
Palliative chemotherapy	1 (1.8%)
Local
CT-guided high-dose-rate brachytherapy	5 (8.9%)
Thermal ablation	2 (3.6%)
Transarterial chemoembolization	8 (14.3%)
Resection
Hemihepatectomy	7 (12.5%)
Time from baseline imaging to TARE evaluation (days)
Mean ± standard deviation	9.91 ± 12.97
Range	0–61
Hepatopulmonary shunt fraction (%)
Mean ± standard deviation	13.11 ± 7.6
Range	2.8–35.97
Treatment after TARE evaluation (n)
1 TARE	24 (42.9%)
2 TARE	12 (21.4%)
≥3 TARE	9 (16.1%)
Additional transarterial chemoembolization	1 (1.8%)
Additional CT-guided high-dose-rate brachytherapy	2 (3.6%)
Additional Sorafenib	3 (5.4%)
No treatment	11 (19.6%)
Follow-up time (months)
Mean ± standard deviation	11.54 ± 16.3
Range	0–115.4
Survival (months)
Patients died during follow-up (n)	18 (32.1%)
Mean time until death ± standard deviation (days)	11.05 ± 6.31
Range (days)	1.18–24.56

Abbreviation: TARE, transarterial radioembolization.

Table 2 Imaging Biomarkers and Tumor Characteristics on Tri-Phasic Contrast-Enhanced CT within six Weeks Before Hepatopulmonary Shunt Evaluation Study

Characteristic	Value
Number of lesions (n)
1	15 (26.8%)
2	1 (1.8%)
3	12 (21.4%)
4	2 (3.6%)
5	2 (3.6%)
6–10	9 (16.1%)
>10–20	4 (7.1%)
>20–30	8 (14.3%)
>30	3 (5.4%)
Diameter [index tumor] (cm)
Mean ± standard deviation	9.67 ± 4.41
Range	2.1–21
Volume [index tumor] (cm^3)
Mean ± standard deviation	404.05 ± 464.24
Range	1.32–2036.44
Diameter [up to five largest lesions] (cm)
Mean ± standard deviation	16.81 ± 7.55
Range	2.1–40.3
Volume [up to five largest lesions] (cm^3)
Mean ± standard deviation	461.05 ± 518.09
Range	1.32–2328.62
Portal vein status (n)
Occlusion
Segmental	16 (28.6%)
Central right/left	18 (32.1%)
Thrombosis
Segmental	3 (5.4%)
Central right/left	21 (37.5%)
Compression
Segmental	16 (28.6%)
Central right/left	15 (26.8%)
Tumor growth pattern [index tumor] (n)
Infiltrative/ill-delineated	22 (39.3%)
Nodular/well-delineated	25 (44.6%)
Cirrhotomimetic	9 (16.1%)
Further imaging biomarkers [index tumor] (n)
Rim arterial phase enhancement	22 (39.3%)
Non-rim arterial phase enhancement	34 (60.7%)
Washout	29 (51.8%)
Capsule	12 (21.4%)
Artery in lesion (arterial/venous)	49 (87.5%) / 31 (55.4%)
Peri-tumoral enhancement	12 (21.4%)
Hypodense halo	12 (21.4%)
Tumor-liver difference	11 (19.6%)
Radiogenomic venous invasion	16 (28.6%)

Table 3 Summary of Regression Models to Predict the Hepatopulmonary Shunt Fraction Including Constants, Significant Variables, Non-Standardized Coefficients, Standardized Coefficients, Standard Error, R², Adjusted R², and F-Statistics

Variable	Non-Standardized Coefficient β (95% Confidence Interval)	Standardized Coefficient	Standard Error ε
Model 1			6.181
Constant	7.049** (1.927–12.172)		2.532
Diameter [index tumor] (cm)	0.560* (0.120–0.999)	0.337*	0.217
Nonrim APHE [index tumor] (no=0; yes=1)	8.137*** (3.476–12.799)	0.545***	2.305
Washout [index tumor] (no=0; yes=1)	−8.171*** (−12.774 – −3.568)	−0.559***	2.276
R^2	0.350
Adjusted R^2	0.300
F (df=3, 39)	7.005**
Model 2			5.970
Constant	7.231** (2.826–11.636)		2.176
Volume [index tumor] (cm^3)	0.005* (0.000–0.009)	0.295*	0.002
Nonrim APHE [index tumor] (no=0; yes=1)	8.677*** (4.058–13.296)	0.581***	2.282
Washout [index tumor] (no=0; yes=1)	−7.763*** (−12.163 – −3.364)	−0.531***	2.173
R^2	0.409
Adjusted R^2	0.347
F (df=4, 38)	6.758***
Model 3			6.638
Constant	7.410** (2.553–12.266)		2.401
Diameter [index tumor] (groups#)	1.582* (0.358–2.807)	0.346*	0.605
Nonrim APHE [index tumor] (no=0; yes=1)	8.318*** (3.655–12.980)	0.557***	2.305
Washout [index tumor] (no=0; yes=1)	−8.597*** (−13.251 – −3.944)	−0.587***	2.301
R^2	0.359
Adjusted R^2	0.312
F (df=3, 41)	7.646***
Model 4			6.226
Constant	7.651** (2.779–12.523)		2.409
Volume [index tumor] (groups†)	1.372* (0.254–2.490)	0.327*	0.553
Nonrim APHE [index tumor] (no=0; yes=1)	8.646*** (3.912–13.380)	0.579***	2.341
Washout [index tumor] (no=0; yes=1)	−8.001*** (−12.612 – −3.390)	−0.327***	2.280
R^2	0.340
Adjusted R^2	0.290
F (df=3, 39)	6.712**

Notes: Significance levels: *p<0.05, **p<0.01, ***p<0.001; groups# refer to a diameter of <5 cm (variable=1), 5–7.5 cm (variable=2), >7.5–10 cm (variable=3), >10–12.5 cm (variable=4), >12.5–15 cm (variable=5), and >15 cm (variable=6); groups† refer to a volume of <50 cm³ (variable=1), 50–100 cm³ (variable=2), >100–250 cm³ (variable=3), >250–500 cm³ (variable=4), >500–1000 cm³ (variable=5), and >1000 cm³ (variable=6).

Abbreviation: APHE, arterial phase enhancement.

Figure 3 Risk scores illustrating the predicted hepatopulmonary shunt fraction using non-invasive biomarkers on contrast-enhanced CT. Models included diameter of index tumor (A) or volume of index tumor (B), index tumor non-rim arterial phase enhancement (APHE), and washout. Recommendations regarding dose reduction and contraindications for the administration of yttrium-90 microspheres are based on the recommendations of the manufacturer Sirtex Medical Limited/S1-Guideline selective intra-arterial radiotherapy (SIRT) of malignant liver tumors of the German Society for Nuclear Medicine.¹⁵

Figure 4 Arterial and venous axial contrast-enhanced CT images displaying hepatocellular carcinoma lesions (two-sided arrows indicate the maximum diameter of the index lesion) of patients recommended for transarterial radioembolization. Hepatopulmonary shunt fraction (HPSF) measured following angiography is shown above each image. Predicted HPSF using the predictive model based on imaging and clinical disease-specific parameters are shown below each image. Examples of non-rim arterial phase enhancement and washout are indicated by solid and dotted arrows, respectively. Recommendations for dose reduction and contraindications for yttrium-90 microspheres administration are color-coded. HPSF displayed in (A and B) resulted in no dose reduction; HPSF displayed in (C and D) resulted in 20% dose reduction; HPSF displayed in (E and F) resulted in 40% dose reduction; HPSF displayed in (G and H) were interpreted as a contraindication for treatment.¹⁵

Müller SP, Antoch G, Lauenstein T, Amthauer H, Haug AR, Bockisch A. DGN Handlungsempfehlung (S1-Leitlinie): Selektive Intraarterielle Radiotherapie (SIRT) Maligner Lebertumoren. Deutsche Gesellschaft für Nuklearmedizin e.V.; 2014.

 Google Scholar