Yttrium-90 Radioembolization Versus Transarterial Embolization for Lung Carcinoid Hepatic Metastasis

Abstract

Objective

To compare the safety and efficacy of transarterial embolization (TAE) and transarterial radioembolization (TARE) in the treatment of lung carcinoid liver metastasis.

Method

This retrospective, single-institution study included 30 patients with lung carcinoid liver metastasis treated with 53 primary embolization procedures (24 TARE; 29 TAE) between 2009 and 2022. Survival analyses were performed using the Kaplan–Meier method. Adverse events were assessed using the Common Terminology Criteria of Adverse Events Version 5.0.

Results

The local tumor progression-free survival (LTPFS) for the entire cohort was 16.2 (95% confidence interval [CI]: 12.6–22.2) months. Factors associated with poor LTFPS for all patients included chromogranin A positivity (hazard ratio [HR]: 10.13, 95% CI: 2.32–44.31, p = 0.002), previous octreotide treatment (HR: 2.37, 95% CI: 1.12–5.01, p = 0.01), and previous everolimus treatment (HR: 2.4, 95% CI: 1.12–5.01, p = 0.02).

TARE was associated with significantly improved LTPFS; 30.57 (95% CI: 12.4–33.4) months compared with TAE 13.9 (95% CI: 10.87–20.13) months, (p = 0.01). Patients on concurrent everolimus at the time of TARE had a significant improvement in LTPFS (HR: 0.25, 95% CI: 0.16–0.41, p < 0.001).

The median overall survival of the entire cohort was 43.5 (95% CI: 20.5–55.9) months; 40.6 (95% CI: 20–49.0) months for TAE versus 43.5 (95% CI: 12.3–61.4) months for TARE (p = 0.49).

Postprocedure adverse events occurred in 28 out of 74 (38%) treatments, most being grade 1 and 2 (n = 24). Severe adverse events (Common Terminology Criteria of Adverse Events grade ≥3) included severe vomiting after TARE (2), acute kidney injury after TAE (1), and bacteremia after TAE (1).

Conclusion

TARE exhibited better local tumor control compared with TAE in patients with lung carcinoid, with no significant difference in overall survival. Both hepatic arterially directed therapies exhibited acceptable safety profiles.

Introduction

Lung carcinoid tumors are rare neuroendocrine epithelial tumors that represent around 2% of all lung malignancies.^1,2 Lung carcinoid tumors can be classified as low-grade typical carcinoid tumors to intermediate grade atypical tumors.³ Although lung carcinoid tumors are rare malignancies with generally favorable survival, the prevalence of this disease is increasing.¹ Resection is the preferred treatment for the primary tumor; however, there remains limited data guiding treatment of metastatic or unresectable lung carcinoids.⁴ These tumors, particularly atypical carcinoids, reveal a higher rate of metastasis to distant organs, with the most common sites being bone and liver.²

The National Comprehensive Cancer Network (NCCN) guidelines for management of neuroendocrine tumors (NET) of the bronchopulmonary system state that patients with metastatic disease who are asymptomatic, low-grade, and low volume can be observed or treated with a sandostatin analog. Sandostatin analogs play a meaningful role in the management of NET and are primarily used for symptom control, with efficacy in slowing tumor growth by binding somatostatin receptors and modulating intracellular signaling pathways. For those patients with clinically significant tumor burden, evidence of progression, symptomatic disease, intermediate-grade or atypical carcinoid- treatment with everolimus or a sandostatin analog is preferred. In addition, the NCCN specifies that for patients with liver-dominant disease, liver-directed therapy should be used.⁵ For patients with lung carcinoid that has metastasized to the liver, locoregional therapies including transarterial chemoembolization, transarterial embolization (TAE), and yttrium-90 (Y90) transarterial radioembolization (TARE), may offer control of liver-dominant disease; however, there is limited data evaluating the efficacy of liver-directed therapy for lung carcinoids and no study comparing embolization modalities for this disease.⁶ The purpose of this study is to evaluate and compare the safety and efficacy of TAE and TARE in treating liver-dominant lung carcinoid metastasis.

Materials and Methods

The study was conducted with the approval of the hospital’s institutional review board; informed consent was waived. This retrospective analysis included all patients with primary lung carcinoid and hepatic metastasis who underwent either TARE or TAE between November 2009 and December 2022. Data was collected using Research Electronic Data Capture. Patient demographics, treatment history, radiology reports, pathology reports, and images were collected from the electronic medical records and picture archiving and communication system. When available, molecular diagnostic pathology was collected from the institution’s Integrated Mutation Profiling of Actionable Cancer Target testing.^7,8

All patients were evaluated by an interventional radiologist before embolization. The decision to proceed with liver-directed therapy was made in conjunction with the medical oncologist and interventional radiologist. Indications for disease treatment included carcinoid syndrome, progression of disease, or bulk symptoms. Treated patients had an Eastern Cooperative Oncology Group performance status of less than or equal to 2 and a total bilirubin level of less than or equal to 2 mg/dL. Radioembolizations were performed with resin (SIR-Spheres; Sirtex, Woburn, MA) or glass microspheres (Therasphere; Boston Scientific, Marlborough, MA), as previously described.⁹ The decision to treat using resin or glass microspheres was made at the discretion of the treating physician. TAE were performed as previously described using polymer-based Embospheres (40–120 μm, 100–300 μm, 300–500 μm, 500–700 μm [Merit Medical, South Jordan, UT]) and Polyvinyl alcohol foam (100 μm, 300 μm [Cook Medical, Bloomington, IN]).¹⁰ TAE procedures were performed with the intent of targeting all intrahepatic disease; for patients with diffuse bilobar disease, treatments were divided into two separate treatment sessions. For patients with more focal or localized disease, selective treatment of involved segments was performed. All patients received 250 μg of subcutaneous octreotide within two hours of treatment.

Y90 single-photon emission computed tomography/computed tomography or positron emission/computed tomography scans were acquired after each radioembolization session. Subsequently, MIM Sureplan software v.6.9 (MIM Software, Cleveland, OH) was used to delineate a volumetric region of interest encompassing the five largest tumors in each treatment territory. The dose to the five largest tumors was then determined using dose volume histograms generated by the software (Supplementary Fig. 1). The maximum tumor dose refers to the highest dose recorded in a single voxel within the tumor region, whereas the mean dose represents the average dose across all voxels in that area. Radiation activity and dose delivered to the liver and mean dose across all sites treated were determined according to the medical internal radiation dose model for glass microspheres and the body surface area model for resin microspheres, both in accordance with the manufacturer’s recommendations for each device. Summary statistics of the absorbed doses to the target treatment area and the lung, lung shunt fraction, treated tumor volume, and percentage of liver treated are included in Supplementary Table 1.

The cohort included 30 patients who underwent 74 overall treatments (53 primary embolization procedures; 21 retreatments). Primary embolization treatments included 24 TAREs and 29 TAEs; five TAREs and 16 TAEs were retreatments. In a previous study, the authors reported on 13 patients included in this study; the previous report evaluated the response and safety of TARE in a mixed lung primary population.⁹ The current study expands on this by including a larger number of patients and compares TARE to TAE in patients with metastatic lung carcinoid. Patients were followed by the interventional clinic with laboratory tests and imaging 4 to 8 weeks after treatment and then at three-month intervals thereafter. Posttreatment laboratory tests were compared with baseline values. Local progression was defined as progression of tumors in the treated segment(s) of the liver, as determined by Response Evaluation Criteria in Solid Tumors, version 1.1.¹¹ Adverse events were graded according to Common Terminology Criteria for Adverse Events version 5.0.¹²

Statistical Analysis

Patient characteristics were summarized using frequency and percent for categorical variables and median and interquartile range for continuous variables. Survival outcomes were calculated from the date of treatment to confirmation of disease progression, death, or last available follow-up. Local tumor progression-free survival (LTPFS), PFS, and overall survival (OS) were evaluated using Kaplan–Meier methodology. LTPFS was evaluated from the time of the primary treatment until the first occurrence of LTP (primary LTPFS) in the treated segment(s). Repeat treatments of the same lesion for local tumor control were not included in the calculation of LTPFS. Patients without progression, who were alive at the time of analysis, were censored to the date of the last available imaging. PFS was evaluated from the time of primary treatment until the first radiographic evidence of new or progressive disease at any site. A cluster term was incorporated to account for patients with multiple treatments. Time to ascites development was also estimated by Kaplan–Meier methodology, from the time of the first treatment. The ability to conduct multivariable analysis was limited by the small sample sizes; multivariable analysis could not be run for all groupwide analyses and within subgroups, as the number of events was insufficient. Differences between groups were estimated using the Cox proportional hazards model. Laboratory toxicities were reported for all procedures (primary and secondary). Data analysis was conducted with Stata, version 17 (StataCorp, College Station, TX).

Results

There was no significant difference in follow–up times between the TAE and TARE groups (p = 0.20); the median follow–up time was not reached for the TAE group (95% confidence interval [CI]: 27–not reached [NR]) and was 85 months for the TARE group (95% CI: 21–NR). Patient and tumor characteristics, along with comparisons between the TAE and TARE groups, are detailed in Table 1. Patients undergoing TAE were more likely to be on concurrent octreotide long-acting release (LAR) treatment at the time of procedure; 11 of 16 (68%) patients in the TAE group versus 1 of 14 (7%) in the TARE group (p = 0.001). Patients treated with TARE had tumors with a higher mitotic count; mean 1.57 (SD: 0.14) versus mean 1.13 (SD: 0.09) for the TAE group (p = 0.01). More patients with functional NETs were present in the TAE group (11/16, 68.75%) than the TARE group (0/14, 0%) (p = 0.045).

Table 1.

Patient Demographics, TAE vs TARE

Variable	TAE (n = 16)	TARE (n = 14)	p Value
Patients
Sex (M, F)	8, 8	6, 8	0.49
Age (median y)	65.5 (IQR: 60.5–71.5)	63 (48–78)	0.9
BMI (median)	25.55 (IQR: 22.2–29.2)	28.5 (24.5–31.7)	0.33
Primary treatments (n)	29	24
Retreatments (n)	16	5
Total (n)	45	29
Chromogranin A
Positive	12	11
Negative	0	1
n/ab	4	2	0.66
Tumor differentiation (n)			1
Well-differentiated	12 (75%)	11 (79%)
Moderately differentiated	1 (6%)	0
Poorly differentiated	1 (6%)	0
n/ab	2 (13%)	3 (21%)
Ki-67 Grade (n, %)			0.26
1	1 (6%)	3 (22%)
2	9 (56%)	8 (57%)
3p	4 (25%)	1 (7%)
n/ab	2 (13%)	2 (14%)
Extrahepatic disease (n)
Lymph node	5 (31%)	5 (36%)	1
Bone	5 (31%)	6 (43%)	0.71
Size of largest tumor (median size, cm)	4.6 (IQR: 3.05–6.85)	5 (IQR: 3.1–5.8)	0.87
Functional Status (n)			0.045
Functional	5 (31%)	0
Nonfunctional	11 (69%)	14 (100%)
Mitotic Count (Mean, SD)	1.13 (.09)	1.57 (.14)	0.008
Previous octreotide LAR (n)	11 (68%)	4 (28%)	0.07
Existing octreotide LAR (n)	11 (68%)	1 (7%)	0.001
Previous everolimus	3 (19%)	3 (21%)	0.85
Existing everolimus	2 (13%)	2 (14%)	0.89
Presence of tumors			0.25
Unilobar	1 (6%)	3 (21%)
Bilobar	15 (94%)	11 (79%)
Number of tumors			0.49
<5	2 (13%)	2 (14%)
≥5	14 (88%)	12 (86%)
Tumor burden			0.11
<50%	8 (50%)	11 (79%)
>50%	8 (50%)	3 (21%)
Lung carcinoid			0.63
Typical	2 (13%)	2 (14%)
Atypical	8 (50%)	9 (64%)
Indeterminatea	6 (38%)	3 (22%)
Previous treatments, n (%)
Biliary stenting	0	0
Hepatic resection	0	0
Hepatic embolization	0	0
Chemoembolization	0	0
Chemotherapy	9 (56%)	12 (86%)	0.09
Octreotide LAR	11 (69%)	4 (36%)	0.07
PRRT	3 (19%)	0
Everolimus	3 (19%)	3 (21%)
Sunitinib-Sutent	1 (6%)	1 (7%)
Existing treatments, n (%)
Biliary stenting	0	0
Hepatic resection	0	0
Hepatic embolization	0	0
Chemoembolization	0	0
Chemotherapy	3 (19%)	4 (36%)
Octreotide	11 (69%)	1 (7%)
PRRT	0	0
Everolimus	2 (13%)	2 (14%)
Sunitinib-Sutent	0	0
Genetic mutations, n (%)
TP53	0	1 (7%)
RB1	4 (25%)	1 (7%)
KRAS	0	1 (7%)
BRAF	0	0
DAXX	0	0
MEN1	2 (13%)	1 (7%)
ATRX	0	0
WT	4 (25%)	1 (7%)
n/ab	7 (44%)	10 (71%)

BMI, body mass index; F, female; IQR, interquartile range; LAR, long-acting release; M, male; n/a, not applicable; PRRT, peptide receptor radionuclide therapy; TAE, transarterial embolization; TARE, transarterial radioembolization; WT, wild-type.

^aDistinction between atypical and typical could not be made on pathologic analysis.

^bVariables were not available.

Local Tumor PFS

The LTPFS for patients across both groups was 16.2 months (95% CI: 12.6–22.2 mo) (Supplementary Fig. 2). Significant factors associated with poor LTPFS for all patients included chromogranin A positivity (HR: 10.13, 95% CI: 2.32–44.31, p = 0.002); previous octreotide treatment (HR: 2.37, 95% CI: 1.12–5.01, p = 0.01); and previous everolimus treatment (HR: 2.4, 95% CI: 1.12–5.01, p = 0.02). All factors retained statistical significance in multivariable analysis; chromogranin A positivity (HR: 2.10, 95% CI: 1.29–3.42, p = 0.003); previous octreotide treatment (HR: 2.74, 95% CI: 1.36–5.54, p = 0.005); and previous everolimus treatment (HR: 3.55, 95% CI: 1.46–8.62, p = 0.005). The impact of typical versus atypical lung carcinoid status on LTPFS is detailed in Supplementary Table 2.

LTPFS after TAE was 13.9 (95% CI: 10.87–20.13) months versus 30.57 (95% CI: 12.4–33.4) months after TARE (p = 0.01) (Fig. 1). For those patients treated with TAE, factors predictive of shorter LTPFS included unilobar disease distribution compared with bilobar (HR: 2.37, 95% CI: 1.26–4.47, p = 0.01). For patients treated with TARE, predictive factors for longer LTPFS included concurrent treatment with everolimus (HR: 0.25, 95% CI: 0.16–0.41, p <0.001). Additional factors associated with LTPFS are highlighted in Table 2.

Figure 1.

LTP for TAE and TARE. LTP, local tumor progression-free survival; TAE, transarterial embolization; TARE, transarterial radioembolization.

Table 2.

Factors Associated With LTPFS

Variable HR (95% CI)		p Value
Age
Entire group	1.01 (0.97–1.05)	0.59
TAE	0.99 (0.96–1.03)	0.72
TARE	1.03 (0.97–1.10)	0.33
Sex
Entire group	1.38 (0.62–3.10)	0.43
TAE	1.25 (0.46–3.40)	0.44
TARE	1.77 (0.44–7.10)	0.80
Chromogranin A
Entire group	10.1 (2.32–44.32)	0.002
TAEa	1	--
TARE	6.8 (0.94–48.67)	0.06
Tumor differentiation
Well-differentiated	Reference
Moderately differentiated
Entire group	1.35 (0.71–2.58)	0.36
TAE	0.96 (0.44–2.06)	0.91
TAREa	1 (––)
Poorly differentiated
Entire groupa	1 (––)	--
TAEa	1 (––)	--
TAREa	1 (––)	--
Tumor grade (On the basis of Ki-67%)
Grade 1	Reference
Grade 2
Entire group	2.06 (0.62–6.824)	0.24
TAE	1.53 (0.71–3.29)	0.28
TARE	1.26 (0.18–8.73)	0.82
Grade 3
Entire group	3.19 (0.83–12.20)	0.09
TAE	1.70 (0.67–4.29)	0.26
TARE	3.93 (0.30–52.01)	0.30
Previous octreotide LAR
Entire group	2.4 (1.20–4.65)	0.01
TAE	1.4 (0.60–3.2)	0.44
TARE	3.3 (0.92–11.7)	0.07
Concurrent octreotide LAR
Entire group	1.47 (0.68–3.14)	0.33
TAE	1.11 (0.42–2.89)	0.84
TARE	0.81 (0.37–1.77)	0.60
Previous everolimus
Entire group	2.4 (1.12–5.01)	0.02
TAE	1.8 (0.77–4.3)	0.17
TARE	4.5 (0.70–28.3)	0.11
Concurrent everolimus
Entire group	0.47 (0.15–1.48)	0.20
TAE	1.44 (0.59–3.51)	0.42
TARE	0.26 (0.16–.413)	0.001
Presence of unilobar vs bilobar disease
Entire group	1.56 (0.63–3.89)	0.34
TAE	2.37 (1.26–4.47)	0.008
TARE	1.03 (0.28–3.78)	0.97
Presence of <5 vs ≥5 tumors
Entire group	1.48 (0.68–3.24)	0.32
TAE	1.32 (0.39–4.42)	0.65
TARE	1.25 (0.38–4.11)	0.72
Tumor burden <50% vs ≥50%
Entire group	1.05 (0.52–2.14)	0.88
TAE	0.49 (0.87–1.06)	0.15
TARE	1.57e–15 (6.78e–17 – 3.65e–14)	< 0.001
Size of largest tumor
Entire group	1.004 (0.903–1.12)	0.94
TAE	0.96 (0.868–1.07)	0.44
TARE	0.98 (0.709–1.35)	0.89
Typical vs atypical carcinoid
Entire group	1.97 (0.76–5.08)	0.16
TAE	0.74 (0.26–2.14)	0.58
TARE	3.18 (0.84–12.06)	0.09

CI, confidence interval; HR, hazard ratio; LAR, long-acting release; LTPFS, local tumor progression-free survival; TAE, transarterial embolization; TARE, transarterial radioembolization.

^aCoefficients could not be estimated because of collinearity (covariate did not vary within the associated risk set).

Progression-Free Survival

The median PFS for patients across both groups was 8.37 (95% CI 5.33–13.9) months (Supplementary Fig. 3). Tumor differentiation was a significant predictor of PFS for all patients, with moderately differentiated tumors showing improved PFS when compared with the reference of well-differentiated tumors (HR: 0.39, 95% CI 0.15–0.98, p = 0.045). Poorly differentiated tumors had a worse PFS compared with the reference of well-differentiated tumors (HR 2.91, 95% CI: 1.58–5.36, p = 0.001).

PFS after TAE was 5.33 (96% CI: 4.2–8.5) months versus 15.9 (95% CI: 6.57–22.23) months after TARE (p = 0.128) (Fig. 2). For those patients treated with TAE, tumor differentiation was a significant predictor of PFS, with moderately differentiated tumors being associated with longer PFS compared with the reference of well-differentiated tumors (HR 0.35, 95% CI: 0.16–0.78, p = 0.01). Tumor grade was also a significant predictor of PFS; grade 2 tumors had a significantly shorter PFS compared with grade 1 (HR 2.96, 95% CI: 1.18–7.44, p = 0.02); grade 3 tumors also had a shorter PFS compared with grade 1 (HR 10.87, 95% CI: 2.95–40.05, p ≤ 0.001). For patients treated with TARE, predictive factors for longer PFS included previous treatment with octreotide LAR (HR 0.14, 95% CI: 0.03–0.75, p = 0.02) and concurrent treatment with octreotide LAR (HR 0.41, 95% CI: 0.20–0.87, p = 0.02); whereas concurrent everolimus treatment was predictive for shorter PFS (HR 7.99, 95% CI: 2.45–26.12, p = 0.001). Additional factors associated with PFS are highlighted in Supplementary Table 3.

Figure 2.

PFS for TAE and TARE. PFS, progression-free survival; TAE, transarterial embolization; TARE, transarterial radioembolization.

Overall Survival

The OS of all treated patients was 43.5 (95% CI: 20.5–55.9) months (Supplementary Fig. 4). The OS times were similar between TAE and TARE groups: 40.6 (95% CI: 20.5–49.0) months versus 43.5 (95% CI: 12.3–61.4) months, respectively (p = 0.49) (Fig. 3). Previous everolimus treatment was associated with significantly poorer OS for the entire cohort (HR: 4.62, 95% CI: 1.39–15.37, p = 0.013). Significance of previous everolimus treatment and poorer OS were also maintained on multivariable analysis (HR: 5.43, 95% CI: 1.06–27.69, p = 0.042). Patients with previous everolimus treatment and TAE had significantly worse OS compared with those without everolimus treatment (HR: 7.72, 95% CI: 1.28–46.47, p = 0.026). Previous everolimus in patients treated with TARE did not impact OS (HR: 2.27, 95% CI: 0.41–12.59, p = 0.35). Complete results of univariate analysis of factors affecting OS are presented in Table 3.

Figure 3.

OS for TAE and TARE. OS, overall survival; TAE, transarterial embolization; TARE, transarterial radioembolization.

Table 3.

Factors Associated With OS

Variable HR (95% CI)		p Value
Age
Entire group	1.00 (0.96–1.04)	0.87
TAE	1.02 (0.96–1.09)	0.44
TARE	0.98 (0.92–1.04)	0.44
Sex
Entire group	2.0 (0.78–5.13)	0.15
TAE	3.46 (0.63–18.90)	0.15
TARE	1.91 (0.47–7.84)	0.37
Chromogranin A positivity
Entire group	1.63 (0.97–2.73)	0.07
TAE	1.42 (0.70–2.89)	0.33
TARE	1.67 (0.76–3.68)	0.21
Tumor Differentiation
Well-differentiated	Reference
Moderately differentiated
Entire group	2.61 (0.32–21.32)	0.37
TAE	3.09 (0.32–30.22)	0.33
TAREa	–––
Poorly differentiated
Entire group	3.20e+16 (0–.)
TAE	1.21e+16 (0–.)
TAREa	––
Grade 1 (On the basis of Ki-67%)	Reference
Grade 2 tumor
Entire group	4.17 (0.90–19.25)	0.07
TAE	1.90 (0.21–17.86)	0.58
TARE	5.17 (0.62–43.33)	0.13
Grade 3 tumor
Entire group	3.61 (0.53–24.33)	0.19
TAE	2.83 (0.23–34.56)	0.42
TAREa	2.22e–19 (–––)	----
Previous octreotide LAR
Entire group	1.11 (0.48–2.57)	0.81
TAE	0.42 (0.08–2.15)	0.30
TARE	1.16 (0.30–4.44)	0.83
Concurrent octreotide LAR
Entire group	0.94 (0.40–2.22)	0.90
TAE	0.48 (0.09–2.46)	0.38
TARE	0.56 (0.07–4.58)	0.59
Previous everolimus
Entire group	4.62 (1.39–15.37)	0.01
TAE	7.72 (1.28–46.47)	0.03
TARE	2.28 (0.41–12.59)	0.35
Concurrent everolimus
Entire group	0.73 (0.21–2.55)	0.63
TAE	1.41 (0.16–12.63)	0.76
TARE	0.73 (0.15–3.50)	0.70
Presence of unilobar vs bilobar disease
Entire group	1.60 (0.47–5.52)	0.45
TAE	0.57 (0.06–4.96)	0.61
TARE	2.09 (0.45–9.70)	0.35
Presence of <5 vs ≥5 tumors
Entire group	2.64 (0.61–11.48)	0.20
TAE	0.91 (0.11–7.78)	0.93
TARE	3.80 (0.48–30.24)	0.21
Tumor burden <50% vs ≥50%
Entire group	2.19 (0.92–5.19)	0.08
TAE	0.83 (0.18–3.86)	0.81
TARE	19.94 (1.96–202.89)	0.01
Size of largest tumor
Entire group	0.98 (0.83–1.15)	0.79
TAE	0.90 (0.71–1.14)	0.36
TARE	1.00 (0.72–1.39)	0.99
Typical vs atypical carcinoid
Entire group	9.50 (1.16–77.80)	0.036
TAE	1.61 (0.17–15.59)	0.68
TARE	1.24e+15	1

CI, confidence interval; HR, hazard ratio; LAR, long-acting release; OS, overall survival; TAE, transarterial embolization; TARE, transarterial radioembolization.

^aCoefficients could not be estimated because of collinearity (covariate did not vary within the associated risk set).

In the total cohort and in patients who received TARE, median OS for typical carcinoid tumors was significantly longer than for atypical carcinoid tumors. OS for atypical and typical carcinoid subtypes is highlighted in Supplementary Table 4. The significance of improved median OS for typical carcinoid tumors compared with atypical across the entire patient cohort was also maintained on multivariable analysis (HR: 8.93, 95% CI: 1.07–74.35, p = 0.043).

Dosimetry Analysis

Dosimetry summary statistics on the basis of tumor dose and impact on LTPFS and OS are detailed in Supplementary Table 5. Larger tumor volume, as calculated by the largest five tumors, was found to be a significant predictive factor for improved LTPFS (HR: 0.994, 95% CI: 0.988–1, p = 0.05) but not for OS (HR: 0.999, 95% CI: 0.994–1.004, p = 0.73). Of note, the percentage of tumors receiving the intended mean dose of Y90 did not impact LTPFS or OS.

Adverse Events

Postprocedure adverse events requiring hospitalization after the intervention or prolonged primary hospitalization occurred after 28 of 74 (38%) treatments, most being grade 1 and 2 (n = 24/28, 86%). Adverse events are highlighted in Supplementary Table 6. One patient who underwent two separate TARE treatments experienced grade 3 severe nausea and vomiting requiring prolonged hospitalization immediately after both treatments. Two grade 4 adverse events after TAE occurred. One patient was initially admitted for the monitoring of transaminitis, hyperbilirubinemia, and worsening ascites, and received TAE treatment for progression of liver disease after initial improvement of liver function. This patient remained hospitalized for acute kidney injury in the context of increased creatine (1.4 mg/dL before treatment to 1.8 mg/dL within 48 h after TAE). The patient was managed on intravenous hydration, and diuretics were held; the creatine stabilized, and the patient was discharged 16 days after treatment. The acute kidney injury was hypothesized to be related to exacerbated hepatorenal syndrome or contrast-induced nephropathy. The second patient with a grade 4 adverse event developed bacteremia 12 days after TAE treatment. The patient had no previous biliary interventions and was hospitalized at an outside facility for 7 days and discharged on 1 week of metronidazole and levofloxacin. Information regarding this patient’s outside hospital course was limited.

Laboratory toxicities, specifically related to liver function, are detailed in Supplementary Table 7. The median time to ascites development for the entire cohort was 46.6 (95% CI: 38.7–59.3) months. Time to development of ascites was not significantly different between the two treatment modalities: the median time in TAE was 46.7 (95% CI 30.0–NR) months versus 42.7 (95% CI: 10.0–60.2) months in TARE (p = 0.77).

Discussion

The prognosis of lung carcinoids is greatly influenced by disease stage, with stage I disease being associated with a 93% 5-year survival rate compared with only 57% for those with stage IV or distant metastatic disease.¹³ In a large review, Trikalinos et al.¹⁴ found that the presence of liver metastases portended a worse prognosis than bone or brain metastases in patients with metastatic NET. Fortunately, for patients with liver-dominant metastatic disease, there are data to suggest that liver-directed therapy may achieve disease control and long-term survival in patients who are both surgical and nonsurgical candidates. In a multi-institutional study, Mayo et al.¹⁵ reviewed outcomes of patients with NET liver metastases treated with surgery and found that the median OS after the first liver resection was more than 125 months. Notably, patients whose hepatic resection was performed with curative intent had a significantly improved OS compared with those treated with palliative intent, 156.9 months versus 77.5 months. For patients with unresectable NET liver metastases, a large review of transarterial chemoembolization and TARE by Egger et al.⁶ reported that both therapies provided a safe and effective means of disease control, with OS being 35.9 months versus 50.1 months, respectively.

Lung carcinoid liver metastases are often multifocal and bilateral at the time of diagnosis, making surgery challenging, if not impossible. For this subset of patients, there remains a need for treatment, which embolization can provide. TAE has been associated with a median OS as long as 80 months, which rivals that seen in the surgical literature for noncurative resection of hepatic metastases.^15,16 Whereas there are numerous studies reporting the efficacy of embolization for NET liver metastases, there is a paucity of data evaluating its efficacy in treating liver metastases from lung carcinoids, in particular.⁹ The OS of the entire cohort in this study was 42 months. One significant predictor of improved OS is typical compared with atypical carcinoid. This echoes the available literature, which has exhibited longer survival time in patients with typical carcinoid compared with atypical.^17,18 This finding may be rather intuitive given the more bland histological features of typical carcinoids, which have fewer than two mitoses per 2 mm² and lack any evidence of necrosis, versus the higher mitotic count and presence of necrosis seen in atypical tumors.¹⁹

The LTPFS across the entire cohort in this study was 16.2 months. Negative prognosticators of LTPFS across all patients treated with liver-directed therapy included presence of chromogranin A on liver biopsy; presence of more than one liver metastasis; and previous treatment with either everolimus or octreotide. However, treatment with octreotide was notably a predictor of improved PFS in patients who received TARE, whereas treatment with everolimus was a negative predictor. Elevations in chromogranin A levels are correlated with tumor burden and presence of metastatic disease, which likely explains its negative prognostication in the current study.²⁰ Larger tumor burden, as denoted by more than one metastatic lesion in the current study, has similarly been found to predict worse outcomes in patients with NETs undergoing intraarterial therapy.²¹ In addition, the negative association between LTPFS and receipt of previous everolimus or octreotide therapy is speculated to be a result of previous exposure, reflecting a more heavily treated patient subset. The positive impact of octreotide on PFS may speak to the impact systemic therapies have on extrahepatic disease that is not addressed by liver-directed therapies.

The present study found that treatment with TARE resulted in a significantly improved LTPFS compared with TAE—30.6 months compared with 13.9 months, respectively. However, it should be noted that these patient groups were statistically different; the TARE cohort had a higher mitotic proliferation index compared with the TAE group, was less likely to have functional disease, and was less likely to be on adjuvant octreotide during their embolization treatment. Studies have found that a higher mitotic index is associated with a worse prognosis; in lung carcinoids, higher mitotic indices are associated with atypical carcinoids, which represent more aggressive disease.^2,22 In addition, there were significantly more functional tumors in the TAE group compared with the TARE group. Whereas there is clearly a difference in patient symptomatology between these groups because of the hormonal excess seen with functional disease, no study has reported a statistically significant improvement in survival in patients with nonfunctional neuroendocrine cancers.23, 24, 25, 26 In fact, there are two studies that have reported improved survival in patients with functional NET compared with their nonfunctional counterparts.^25,26 The greater incidence of concurrent octreotide treatment in the TAE group is likely secondary to those patients having a significantly higher rate of functional disease.

In the present study, TARE with existing everolimus treatment was associated with significantly improved LTPFS. The randomized, double-blind RADIANT-4 trial reported improved PFS with everolimus in patients with progressive lung and gastrointestinal NETs compared with a placebo; everolimus was associated with a 52% reduction in the estimated risk of progression or death.²⁷ Everolimus treatment arrests cell division and has been postulated to increase cytotoxicity of radiation therapy by inhibiting DNA repair, and potentiating antiangiogenic effects of embolic therapy. In a phase 1b study by Kim et al.²⁸ escalating everolimus dosage with pasireotide and concurrent TARE treatment for NET was found to have minimal, reversible liver toxicity, and PFS was more than 18 months. The current study also revealed a significant benefit in PFS in those patients who underwent TARE and had past or current use of octreotide. Somatostatin analogs alone have been found to promote stabilization in patients with well-differentiated NETS; these studies included lung neuroendocrine tumor patients.29, 30, 31 Sullivan et al.³¹ retrospectively evaluated the impact of somatostatin analogs in patients with progressive, metastatic pulmonary carcinoids and found that 77% of patients had stable disease, with a median duration of treatment of 13.7 months. Findings in the present study suggest that there may be a synergistic effect between everolimus and octreotide and TARE, allowing for improved therapeutic effect.

As previously mentioned, no difference in PFS or OS was observed between the two treatments. The lack of a survival advantage, despite the impressive difference in LTPFS, is thought to be owing to patient selection and the small sample size. At the authors’ institution, TAE is primarily used as a first-line treatment modality because of concerns of radiation induced liver damage. Another consideration is that the long survival associated with metastatic NETs is likely confounded by the adjunctive and systemic therapies performed. The long time to event (OS and PFS) likely requires a larger sample size than that seen in this small retrospective study; perhaps an impact on survival could be ascertained by a larger number of participants. Whereas no previous studies compared embolization therapies for lung carcinoid liver metastases, the limited data in the literature regarding radioembolization and bland embolization for gastrointestinal NETs suggest equivocal outcomes.³² The results of this study suggest superiority in LTPFS after radioembolization in patients with lung carcinoid tumors and liver metastases; however, it should be noted that there are differences in patient baseline characteristics. The NCCN guidelines recommend caution with TARE because of the risk of radioembolization-induced chronic hepatotoxicity in long-term survivors that may require repeated treatments.⁵ Radioembolization-induced chronic hepatotoxicity and radioembolization-induced liver disease are rare adverse events after TARE in which patients present with jaundice, ascites, hepatic laboratory derangements, hepatic necrosis, hepatic failure, portal hypertension, or portal vein thrombosis.³³ Wong et al.³⁴ evaluated long-term outcomes in patients with NET who had received TARE and found that ascites developed in 5% of patients in the absence of progressive hepatic disease, and the median time to development of new ascites was only 5.5 months after treatment. The present investigation found that liver toxicity, as represented by the time to development of ascites, was not significantly different between TAE and TARE (46.7 mo versus 42.7 mo, respectively). Whereas the mean time to ascites development in this study was quite long, the time to ascites onset in the literature is variable.

Hepatic laboratory values were also analyzed in the current study. Analysis of liver function after treatment, compared with baseline levels, found that the greatest changes in alanine aminotransferase, aspartate aminotransferase, and alkaline phosphatase happened within the first 6 months after treatment in both the TAE and TARE groups. This is likely because of immediate postintervention liver injury from ischemia or radiation. Notably, median values of alanine aminotransferase, aspartate aminotransferase, and alkaline phosphatase all normalized 12 months after treatment. The current study suggests that short-term laboratory toxicity and long-term toxicity, as marked by the development of ascites, are equivocal between both TAE and TARE groups. In this study, both TAE and TARE treatments were associated with a few severe adverse events. The most common adverse events that resulted in hospitalization within 30 days of treatment for both therapies included fatigue, abdominal pain, and culture-negative fever, which are known symptoms of postembolization syndrome resulting from ischemic necrosis.³⁵ Two patients developed grade 4 adverse events after TAE: acute kidney injury after treatment and bacteremia. Contrast-related kidney injury is a rare adverse event that has been previously described in the literature for transarterial chemoembolization.³⁶ Bacteremia is also a rare adverse event of hepatic embolization, and clinical symptoms may initially present similarly to postembolization syndrome.³⁷ Special consideration should be taken in patients with worsening fevers and pain despite supportive management after embolization.

Limitations of the study include the small sample size and retrospective nature. The aforementioned factors limited the ability to evaluate heterogeneity of effects across subgroups and to conduct multivariable analyses across all covariates. Whereas survival comparisons between those patients treated with TARE and TAE were performed, these patient groups were notably different, which requires caution when interpreting the results of this study. In addition, survival analysis may have been affected by subsequent treatments utilized postembolization, which is something that was not explored in this study. Another limitation of this study was that the Y90 dosimetry was on the basis of the medical internal radiation dose and body surface area models; however, many practices utilize the newer partition model, which allows for more accurate tumor dosing. Despite these limitations, this study provided real-life evidence of the use of TAE and TARE in the management of patients with metastatic lung carcinoid. The strong signal favoring TARE in terms of sustained local tumor control and LTPFS requires further investigation with dedicated prospective studies.

In conclusion, embolization techniques such as TAE and TARE seem to be safe and viable treatment options in patients with lung carcinoid hepatic metastases. In this study, TARE was associated with a significantly improved LTPFS compared with TAE, and the advantage of TARE was further optimized with synchronous everolimus and octreotide treatment. Further research evaluating TARE and TAE for liver-dominant or progressive disease in patients with lung carcinoid is imperative to further guide optimization of treatment.

CRediT Authorship Contribution Statement

Gavin Yuan: Conceptualization, Writing - original draft, Data curation, Formal analysis.

Marios Platon Dimopoulos: Data Curation, Visualization, Conceptualization, Writing - review & editing.

Elena N. Petre: Conceptualization, Supervision, Formal analysis, Methodology, Writing - review & editing.

Etay Ziv: Methodology, Supervision, Investigation, Writing - review & editing.

Constantinos T. Sofocleous: Supervision, Investigation, Resources, Writing - review & editing.

Lee Rodriguez: Data analysis, Visualization, Writing - review & editing.

Alissa Cooper: Investigation, Writing - review & editing.

Vlasios Sotirchos: Investigation, Writing - review & editing.

Ken Zhao: Investigation, Writing - review & editing.

Adrian Gonzalez Aguirre: Data curation, Investigation, Writing - review & editing.

Erica S. Alexander: Conceptualization, Methodology, Writing - original draft, Supervision, Writing - review & editing.

Disclosure

Dr. Ziv received grant support from the American Association for Cancer Research–Neuroendocrine Tumor Research Foundation (AACR-NETRF), Memorial Sloan Kettering, Druckenmiller, Society of Interventional Radiology, Radiological Society of North America, North American Neuroendocrine Tumor Society, Ethicon, and Novartis. Dr. Sofocleous received grant support from National Institutes of Health/National Cancer Institute (NIH/NCI), Society of Interventional Radiology Foundation, Society of Interventional Oncology, Ethicon Johnson and Johnson (J&J), Varian, Boston Scientific, and Sirtex; is a consultant for Ethicon J&J, Varian, Terumo, Medtronic, and Sirtex; and receives equipment/material support from MIM Software, a GE HealthCare company, (Beachwood, Ohio). Dr. Cooper received grant support from Amgen, Roche, Monte Rosa, Daiichi Sankyo, AbbVie, and Merck; is a consultant for Gilead Sciences, Daiichi Sankyo, Novartis, Boehringer Ingelheim (BI), Regeneron, and Amgen; and received payment or honoraria from MJH Life Sciences, Ideology Health, MedStar Health, PeerDirect, CancerGRACE, and Intellisphere LLC. Dr. Sotirchos is a consultant for Intera Oncology. Dr. Alexander is a consultant for Boston Scientific. Dr. Zhao received grant support from the Society of Interventional Radiology. The remaining authors declare no conflict of interest.

Supplementary Data

Supplementary Figure 1.

Supplementary Figure 2.

Supplementary Figure 3.

Supplementary Figure 4.